This document contains information about a computer engineering course on local area networks taught at King Fahd University of Petroleum & Minerals. It includes the course number, term, instructor details, and covers topics like fixed assignment access schemes, time division multiple access (TDMA), frequency division multiple access (FDMA), and their analysis in terms of throughput and delay. Equations and analysis of different access protocols like TDMA, FDMA, ALOHA, and slotted ALOHA are presented. Performance metrics like throughput, delay, and idealized central control with a single queue are also summarized.
Routing in packet switched networks involves selecting the optimal path between end nodes across the network. Adaptive routing is commonly used, where routing decisions dynamically change in response to network conditions like failures or congestion. Key algorithms for computing optimal paths include Dijkstra's algorithm and Bellman-Ford algorithm, which calculate the shortest or least-cost path between all nodes based on link costs.
The document discusses the air interface of GSM mobile networks. It describes how GSM utilizes a combination of frequency division multiple access (FDMA) and time division multiple access (TDMA) on the air interface. This results in an 8-timeslot frame structure across multiple frequencies. Logical channels like traffic channels and control channels are mapped onto these physical timeslots. Control channels include synchronization, broadcast, and paging channels. Precise timing is required between uplink and downlink transmissions to account for signal propagation delays.
Multistage Implementation of Narrowband LPF by Decimator in Multirate DSP App...iosrjce
Decimator is an important sampling device used for multi-rate signal processing in wireless
communication systems. Multirate systems have traditionally played the important role in compression for
contemporary communication application.In this paper it demonstrated that a multistage implementation of
sampling rate conversion often provides for a more efficient realization, especially when filter specifications are
very tight (e.g., a narrow pass band and narrow transition band) and there are a audio-band of 4kHz bandwidth
and compression by decimator to isolate the frequency component 80Hz.The LPF used by the decimator are
acquire by two different approaches: first is the window method and other is frequency sampling
method,Multistages implementations are used to further reduce the computational load. This approach
drastically reduces the filter order and also reduces computational cost. Here it have reduce the overall
computational complexity at single stage is 50 times and for second stages is near about 9 times reduce by the
decimator factor is 50.
This document provides an overview of the LTE physical layer frame structure and signals for both Frequency Division Duplexing (FDD) and Time Division Duplexing (TDD). It describes the basic time and frequency units used in LTE, the uplink and downlink frame structures, reference signals, and the modulation schemes used for different physical channels. Key aspects covered include resource blocks, OFDM signal generation, uplink SC-FDMA modulation, and the use of multiple antennas and spatial multiplexing in the downlink.
Packet radio protocols allow multiple subscribers to access a shared channel for transmitting data packets. They use contention-based random access techniques like ALOHA. Pure ALOHA protocol has low efficiency due to partial packet collisions. Slotted ALOHA synchronizes transmissions to time slots to prevent partial collisions, improving efficiency. Performance is evaluated using metrics like throughput, which is highest at optimal channel load and drops off above and below this point.
This document summarizes a study on modeling GPRS session time distribution. The study group was asked to construct a model for simultaneous transmission of voice and data on mobile networks and determine equipment needs to provide a required quality of service. The summary develops:
1) A Markov chain model where voice calls have priority over data calls. The model accounts for variable voice, data arrival rates and data call sizes.
2) An initial model where each data call uses one channel. Balance equations are developed to calculate the stationary distribution.
3) Future work is outlined to model variable data call channel usage and develop numerical solutions to understand performance metrics like mean wait times.
Performance Evaluation of Routing Protocols Ankush Mehta
This is a research project based on Performance checking of the Routing Protocols. This Presentation shows the basic knowledge of the Protocols use (AODV, DSDV and DSR) and in the end it shows the Result and Conclusion by comparing the graphs which are generated through out the work.
Static channel allocation uses time division multiplexing and frequency division multiplexing to allocate channels to users. Each user is statically assigned a specific portion of the frequency spectrum or time slot. This method is inefficient because it wastes bandwidth if the number of users is less than the number of portions the spectrum is divided into. It also causes delays for new users who need to wait for channels to become available before using the resource.
Routing in packet switched networks involves selecting the optimal path between end nodes across the network. Adaptive routing is commonly used, where routing decisions dynamically change in response to network conditions like failures or congestion. Key algorithms for computing optimal paths include Dijkstra's algorithm and Bellman-Ford algorithm, which calculate the shortest or least-cost path between all nodes based on link costs.
The document discusses the air interface of GSM mobile networks. It describes how GSM utilizes a combination of frequency division multiple access (FDMA) and time division multiple access (TDMA) on the air interface. This results in an 8-timeslot frame structure across multiple frequencies. Logical channels like traffic channels and control channels are mapped onto these physical timeslots. Control channels include synchronization, broadcast, and paging channels. Precise timing is required between uplink and downlink transmissions to account for signal propagation delays.
Multistage Implementation of Narrowband LPF by Decimator in Multirate DSP App...iosrjce
Decimator is an important sampling device used for multi-rate signal processing in wireless
communication systems. Multirate systems have traditionally played the important role in compression for
contemporary communication application.In this paper it demonstrated that a multistage implementation of
sampling rate conversion often provides for a more efficient realization, especially when filter specifications are
very tight (e.g., a narrow pass band and narrow transition band) and there are a audio-band of 4kHz bandwidth
and compression by decimator to isolate the frequency component 80Hz.The LPF used by the decimator are
acquire by two different approaches: first is the window method and other is frequency sampling
method,Multistages implementations are used to further reduce the computational load. This approach
drastically reduces the filter order and also reduces computational cost. Here it have reduce the overall
computational complexity at single stage is 50 times and for second stages is near about 9 times reduce by the
decimator factor is 50.
This document provides an overview of the LTE physical layer frame structure and signals for both Frequency Division Duplexing (FDD) and Time Division Duplexing (TDD). It describes the basic time and frequency units used in LTE, the uplink and downlink frame structures, reference signals, and the modulation schemes used for different physical channels. Key aspects covered include resource blocks, OFDM signal generation, uplink SC-FDMA modulation, and the use of multiple antennas and spatial multiplexing in the downlink.
Packet radio protocols allow multiple subscribers to access a shared channel for transmitting data packets. They use contention-based random access techniques like ALOHA. Pure ALOHA protocol has low efficiency due to partial packet collisions. Slotted ALOHA synchronizes transmissions to time slots to prevent partial collisions, improving efficiency. Performance is evaluated using metrics like throughput, which is highest at optimal channel load and drops off above and below this point.
This document summarizes a study on modeling GPRS session time distribution. The study group was asked to construct a model for simultaneous transmission of voice and data on mobile networks and determine equipment needs to provide a required quality of service. The summary develops:
1) A Markov chain model where voice calls have priority over data calls. The model accounts for variable voice, data arrival rates and data call sizes.
2) An initial model where each data call uses one channel. Balance equations are developed to calculate the stationary distribution.
3) Future work is outlined to model variable data call channel usage and develop numerical solutions to understand performance metrics like mean wait times.
Performance Evaluation of Routing Protocols Ankush Mehta
This is a research project based on Performance checking of the Routing Protocols. This Presentation shows the basic knowledge of the Protocols use (AODV, DSDV and DSR) and in the end it shows the Result and Conclusion by comparing the graphs which are generated through out the work.
Static channel allocation uses time division multiplexing and frequency division multiplexing to allocate channels to users. Each user is statically assigned a specific portion of the frequency spectrum or time slot. This method is inefficient because it wastes bandwidth if the number of users is less than the number of portions the spectrum is divided into. It also causes delays for new users who need to wait for channels to become available before using the resource.
This document provides step-by-step instructions for performing 3GPP Rel-5 transmitter characteristics, receiver characteristics, and performance tests using the R&S CMU200 instrument. It covers tests such as maximum output power, code domain power accuracy, spectrum emission mask, error vector magnitude, channel quality indicator reporting, and HS-SCCH detection performance. The document also includes *.sav files for recalling predefined test configurations for a UE supporting band I and power class 3.
The document provides an overview of implementing 802.11 wireless networks in the ns-2 network simulator. It discusses:
1) Key components of ns-2 including the event scheduler, network components, and support for wireless extensions.
2) Implementation of the 802.11 physical layer including propagation models and sending/receiving packets.
3) Implementation of the 802.11 MAC layer including states, timers, CSMA/CA, and a capture model.
4) An example of adding Rayleigh fading to analyze its impact on TCP and UDP performance over distance.
This document describes three TCP-aware link layer protocols: Snooping TCP, Wireless TCP, and Delayed DACK. Snooping TCP uses an agent at the base station to snoop and buffer TCP connections, ensuring packets are delivered to the mobile node in order and retransmitting lost packets. Wireless TCP modifies timestamps to compensate for increased round-trip time. Delayed DACK delays acknowledgments to allow time for lost packets to be recovered before triggering retransmissions.
This paper proposes and analyzes the performance of a selection decode-and-forward cooperative free-space optical communication system using adaptive subcarrier quadrature amplitude modulation. The system employs selective relaying to choose the best intermediate node based on channel state information. Novel expressions are derived for outage probability, spectral efficiency, and bit error rate considering Gamma-Gamma atmospheric turbulence fading. Numerical results show that the proposed adaptive system has improved performance compared to non-adaptive systems and all-active relaying schemes.
The document discusses various modes of data transmission including simplex, half-duplex, and full-duplex transmission. It also covers parallel versus serial transmission and describes how parallel transmission uses multiple wires while serial transmission uses just two wires. The document then discusses modulation techniques for data transmission over telephone lines using modems as well as synchronization techniques for digital modulation including asynchronous and synchronous methods.
This document proposes a methodology for incorporating uplink delay constraints into LTE cell planning for smart grid applications. It presents the following:
1) A semi-analytical approach is proposed to evaluate uplink transmission delays considering buffering delays before scheduling, packet transmission/retransmission delays over the air interface, and constraints from smart grid standards.
2) Analytical models are used to estimate buffering delays before scheduling based on queue length and service rate. Packet transmission delays are estimated considering packet segmentation, link adaptation, resource block allocation and retransmissions.
3) A cell planning algorithm is analyzed that incorporates these delay metrics to validate compliance with smart grid delay constraints defined in standards. Path loss, interference
OFDM is a digital multi-carrier modulation technique that divides the available spectrum into multiple orthogonal subcarriers. It allows high spectral efficiency by spacing the carriers to maintain orthogonality even when their spectra overlap. The document provides an intuitive explanation of OFDM using analogies like a shower head vs faucet and multiple smaller trucks vs one large truck. It explains how OFDM provides resistance to interference by spreading data across orthogonal subcarriers rather than a single carrier. The key concept of orthogonality allows the subcarriers to overlap without interference by ensuring the area under one subcarrier's frequency multiplied by another is always zero.
This chapter discusses routing in circuit-switched and packet-switched networks. It covers different routing strategies like static, dynamic, alternate and adaptive routing. It also describes routing algorithms like Dijkstra's algorithm and Bellman-Ford algorithm that are used to determine the optimal path between nodes based on performance criteria like minimum hop count or least cost. These algorithms calculate the shortest paths in a network using information about the topology and link costs.
This document evaluates and compares the performance of seven high-speed TCP congestion control protocols: Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP and YeAH TCP. It first provides background on the need for high-speed congestion control as internet speeds have increased. It then summarizes the algorithms and mechanisms of each protocol. The document aims to simulate and compare the performance of these protocols with multiple flows, in order to determine the best approach for high-speed networks.
The performance of wireless ad hoc networks is impacted significantly by the way TCP reacts to lost packets. TCP was designed specifically for wired, reliable networks; thus, any packet loss is attributed to congestion in the network. This assumption does not hold in wireless networks as most packet loss is due to link failure. In our research we analyzed several implementations of TCP, including TCP Vegas, TCP Feedback, and SACK TCP, by measuring throughput, retransmissions, and duplicate acknowledgements through simulation with ns-2. We discovered that TCP throughput is related to the number of hops in the path, and thus depends on the performance of the underlying routing protocol, which was DSR in our research.
COMPARISON OF HIGH SPEED CONGESTION CONTROL PROTOCOLSIJNSA Journal
Congestion control limits the quantity of information input at a rate less important than that of the transmission one to ensure good performance as well as protect against overload and blocking of the network. Researchers have done a great deal of work on improving congestion control protocols, especially on high speed networks. In this paper, we will be studying the congestion control alongside low and high speed congestion control protocols. We will be also simulating, evaluating, and comparing eight of high speed congestion control protocols : Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP, Compound TCP and YeAH TCP, with multiple flows.
This document discusses multiple access protocols for wireless networks. It begins by describing random access protocols like ALOHA and slotted ALOHA. It then covers controlled access protocols using reservation, polling, and token passing. Finally, it discusses channelization protocols using frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA). Throughout are examples calculating throughput for different access loads and determining minimum frame sizes.
For the recent CMOS feature sizes power dissipation becomes an overriding concerns for VLSI circuit design. We propose a novel approach named tri-state buffer with common data bus which reduces the total power & delay of elastic buffer. The paper presents a design and implementation of tri-state buffer mechanism. This design offers also the advantage of third state (High Impedance state) of tri-state buffer. The proposed elastic buffer design using tri-state buffer is implemented in Cadence tools. The obtained result shows that our design is effective in terms 20.50 % reduction in total power, 89.67% reduction in delay.
DAMA is a technique used to assign satellite channels to users on an as-needed basis. It allows a satellite to communicate with different earth stations simultaneously without interference. With DAMA, communication channels are assigned based on requests from user terminals to a network control system. Once allocated, a channel is reserved for a user's session and not available to others until it is finished. This improves efficiency over systems that permanently allocate channels.
The document summarizes key concepts about data link layer protocols from Chapter 4 of the textbook. It discusses media access control using contention-based and controlled access methods. It also covers error control techniques like error detection using parity checks, checksums, and cyclic redundancy checks. Error correction is achieved using retransmission-based automatic repeat request and forward error correction methods. Asynchronous and synchronous transmission protocols are compared, along with common file transfer protocols for asynchronous lines.
Beginners guide to social media (2010 ALGIM Web Symposium)Jason Dawson
A basic introduction to social media and its uses for New Zealand local government presented at the 2010 ALGIM Web Symposium in Wellington. Includes tips on rules of engagement for social media, what works, what to avoid and who is using it in the public sector.
Summary and highlights from the 2013 New Zealand local government website survey, co-ordinated and produced by the Association of Local Government Information Management (ALGIM)
This document provides step-by-step instructions for performing 3GPP Rel-5 transmitter characteristics, receiver characteristics, and performance tests using the R&S CMU200 instrument. It covers tests such as maximum output power, code domain power accuracy, spectrum emission mask, error vector magnitude, channel quality indicator reporting, and HS-SCCH detection performance. The document also includes *.sav files for recalling predefined test configurations for a UE supporting band I and power class 3.
The document provides an overview of implementing 802.11 wireless networks in the ns-2 network simulator. It discusses:
1) Key components of ns-2 including the event scheduler, network components, and support for wireless extensions.
2) Implementation of the 802.11 physical layer including propagation models and sending/receiving packets.
3) Implementation of the 802.11 MAC layer including states, timers, CSMA/CA, and a capture model.
4) An example of adding Rayleigh fading to analyze its impact on TCP and UDP performance over distance.
This document describes three TCP-aware link layer protocols: Snooping TCP, Wireless TCP, and Delayed DACK. Snooping TCP uses an agent at the base station to snoop and buffer TCP connections, ensuring packets are delivered to the mobile node in order and retransmitting lost packets. Wireless TCP modifies timestamps to compensate for increased round-trip time. Delayed DACK delays acknowledgments to allow time for lost packets to be recovered before triggering retransmissions.
This paper proposes and analyzes the performance of a selection decode-and-forward cooperative free-space optical communication system using adaptive subcarrier quadrature amplitude modulation. The system employs selective relaying to choose the best intermediate node based on channel state information. Novel expressions are derived for outage probability, spectral efficiency, and bit error rate considering Gamma-Gamma atmospheric turbulence fading. Numerical results show that the proposed adaptive system has improved performance compared to non-adaptive systems and all-active relaying schemes.
The document discusses various modes of data transmission including simplex, half-duplex, and full-duplex transmission. It also covers parallel versus serial transmission and describes how parallel transmission uses multiple wires while serial transmission uses just two wires. The document then discusses modulation techniques for data transmission over telephone lines using modems as well as synchronization techniques for digital modulation including asynchronous and synchronous methods.
This document proposes a methodology for incorporating uplink delay constraints into LTE cell planning for smart grid applications. It presents the following:
1) A semi-analytical approach is proposed to evaluate uplink transmission delays considering buffering delays before scheduling, packet transmission/retransmission delays over the air interface, and constraints from smart grid standards.
2) Analytical models are used to estimate buffering delays before scheduling based on queue length and service rate. Packet transmission delays are estimated considering packet segmentation, link adaptation, resource block allocation and retransmissions.
3) A cell planning algorithm is analyzed that incorporates these delay metrics to validate compliance with smart grid delay constraints defined in standards. Path loss, interference
OFDM is a digital multi-carrier modulation technique that divides the available spectrum into multiple orthogonal subcarriers. It allows high spectral efficiency by spacing the carriers to maintain orthogonality even when their spectra overlap. The document provides an intuitive explanation of OFDM using analogies like a shower head vs faucet and multiple smaller trucks vs one large truck. It explains how OFDM provides resistance to interference by spreading data across orthogonal subcarriers rather than a single carrier. The key concept of orthogonality allows the subcarriers to overlap without interference by ensuring the area under one subcarrier's frequency multiplied by another is always zero.
This chapter discusses routing in circuit-switched and packet-switched networks. It covers different routing strategies like static, dynamic, alternate and adaptive routing. It also describes routing algorithms like Dijkstra's algorithm and Bellman-Ford algorithm that are used to determine the optimal path between nodes based on performance criteria like minimum hop count or least cost. These algorithms calculate the shortest paths in a network using information about the topology and link costs.
This document evaluates and compares the performance of seven high-speed TCP congestion control protocols: Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP and YeAH TCP. It first provides background on the need for high-speed congestion control as internet speeds have increased. It then summarizes the algorithms and mechanisms of each protocol. The document aims to simulate and compare the performance of these protocols with multiple flows, in order to determine the best approach for high-speed networks.
The performance of wireless ad hoc networks is impacted significantly by the way TCP reacts to lost packets. TCP was designed specifically for wired, reliable networks; thus, any packet loss is attributed to congestion in the network. This assumption does not hold in wireless networks as most packet loss is due to link failure. In our research we analyzed several implementations of TCP, including TCP Vegas, TCP Feedback, and SACK TCP, by measuring throughput, retransmissions, and duplicate acknowledgements through simulation with ns-2. We discovered that TCP throughput is related to the number of hops in the path, and thus depends on the performance of the underlying routing protocol, which was DSR in our research.
COMPARISON OF HIGH SPEED CONGESTION CONTROL PROTOCOLSIJNSA Journal
Congestion control limits the quantity of information input at a rate less important than that of the transmission one to ensure good performance as well as protect against overload and blocking of the network. Researchers have done a great deal of work on improving congestion control protocols, especially on high speed networks. In this paper, we will be studying the congestion control alongside low and high speed congestion control protocols. We will be also simulating, evaluating, and comparing eight of high speed congestion control protocols : Bic TCP, Cubic TCP, Hamilton TCP, HighSpeed TCP, Illinois TCP, Scalable TCP, Compound TCP and YeAH TCP, with multiple flows.
This document discusses multiple access protocols for wireless networks. It begins by describing random access protocols like ALOHA and slotted ALOHA. It then covers controlled access protocols using reservation, polling, and token passing. Finally, it discusses channelization protocols using frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA). Throughout are examples calculating throughput for different access loads and determining minimum frame sizes.
For the recent CMOS feature sizes power dissipation becomes an overriding concerns for VLSI circuit design. We propose a novel approach named tri-state buffer with common data bus which reduces the total power & delay of elastic buffer. The paper presents a design and implementation of tri-state buffer mechanism. This design offers also the advantage of third state (High Impedance state) of tri-state buffer. The proposed elastic buffer design using tri-state buffer is implemented in Cadence tools. The obtained result shows that our design is effective in terms 20.50 % reduction in total power, 89.67% reduction in delay.
DAMA is a technique used to assign satellite channels to users on an as-needed basis. It allows a satellite to communicate with different earth stations simultaneously without interference. With DAMA, communication channels are assigned based on requests from user terminals to a network control system. Once allocated, a channel is reserved for a user's session and not available to others until it is finished. This improves efficiency over systems that permanently allocate channels.
The document summarizes key concepts about data link layer protocols from Chapter 4 of the textbook. It discusses media access control using contention-based and controlled access methods. It also covers error control techniques like error detection using parity checks, checksums, and cyclic redundancy checks. Error correction is achieved using retransmission-based automatic repeat request and forward error correction methods. Asynchronous and synchronous transmission protocols are compared, along with common file transfer protocols for asynchronous lines.
Beginners guide to social media (2010 ALGIM Web Symposium)Jason Dawson
A basic introduction to social media and its uses for New Zealand local government presented at the 2010 ALGIM Web Symposium in Wellington. Includes tips on rules of engagement for social media, what works, what to avoid and who is using it in the public sector.
Summary and highlights from the 2013 New Zealand local government website survey, co-ordinated and produced by the Association of Local Government Information Management (ALGIM)
Este documento presenta los resultados estadísticos de cinco agencias de viajes. Las tablas muestran datos como el número de clientes, ingresos y ganancias de cada agencia para realizar un análisis comparativo.
Social media: delivering on expectationsJason Dawson
Workshop presentation from ALGIM 2011 Local Govt Web Symposium, 2 May 2011. Covers social media policy, monitoring social media, tools to manage social media and examples during crisis management (including Christchurch earthquake).
1) Multiplexing is a technique that combines multiple message signals into a composite signal for transmission over a common channel. The two main types are time-division multiplexing (TDM) used in digital transmission, and frequency-division multiplexing (FDM) used in analogue transmission.
2) In TDM, the time of one channel is divided among multiple users so each appears to have the full channel for a fraction of the total time. In FDM, the frequency of one channel is divided among users so multiple transmissions can occur simultaneously.
3) Modulation involves changing a characteristic of a high-frequency carrier signal based on the instantaneous amplitude of an information signal. This allows multiple signals to be
Spread spectrum is a communication technique that spreads a narrowband communication signal over a wide range of frequencies for transmission then de-spreads it into the original data bandwidth at the receive.
Spread spectrum communications and CDMAHossam Zein
This document discusses spread spectrum communications and CDMA. It provides an overview of spread spectrum techniques including direct sequence spread spectrum (DS/SS), frequency hopping, time hopping, and hybrid techniques. It explains that CDMA uses unique codes to allow multiple access. The document also discusses code properties like autocorrelation and cross-correlation, and code families including maximal length sequences, Gold codes, and Kasami codes that have good correlation properties for CDMA. It notes that correlation properties may be less important with more advanced receivers.
The document discusses multiple access techniques for satellite communications. It describes frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA). FDMA divides the available radio spectrum into narrow frequency channels. TDMA divides each radio channel into time slots. CDMA allows all users to access the full bandwidth at all times by using orthogonal spreading codes. The document provides examples of these different multiple access techniques and compares their approaches.
OFDM is a high-speed wireless transmission technology that divides the available spectrum into multiple orthogonal subcarriers. It is implemented as OFDMA to support multi-user communication. OFDM provides advantages over single carrier transmission by combating inter-symbol interference and frequency selective fading. It works by encoding data over multiple carrier frequencies, with spacing between carriers chosen so that the carriers are orthogonal to each other. This allows high data rates without overlapping signals at a receiver.
This document discusses multiple access protocols used to coordinate access to shared broadcast channels. It describes various channel partitioning protocols like TDMA and FDMA that divide channels by time or frequency. Random access protocols like ALOHA and CSMA are also covered, which allow nodes to transmit randomly and detect collisions. CSMA/CD improves on CSMA by allowing nodes to detect collisions quickly and abort transmissions. Taking-turns protocols pass control of the channel between nodes either through polling or token passing. The document provides examples and compares the efficiency of different multiple access protocols.
This document discusses multiple-input multiple-output (MIMO) systems. It begins by outlining the motivations and aspirations for developing MIMO, including achieving high data rates near 1 Gbps while maintaining quality of service. It then covers MIMO system modeling and capacity studies. Different MIMO designs are presented that aim to achieve spatial multiplexing gain or diversity gain. Practical MIMO systems and architectures like V-BLAST are described. Networking applications of MIMO including MAC protocols are also discussed.
The document discusses multiple access protocols used to coordinate access to shared broadcast channels in wireless and wired local area networks. It describes various channel partitioning, random access, and taking-turns protocols including ALOHA, CSMA, TDMA, FDMA, and token passing. Specifically, it provides details on the operation of slotted ALOHA, CSMA, and CSMA/CD protocols, comparing their efficiencies and ability to handle collisions. It also briefly discusses token ring and FDDI local area network technologies.
The document discusses multiple access protocols used to coordinate access to shared broadcast channels in wireless and wired local area networks. It describes various channel partitioning, random access, and taking-turns protocols such as TDMA, FDMA, CSMA, ALOHA, and token passing. These protocols aim to prevent collisions and allow efficient use of the broadcast channel while accommodating transmission from multiple nodes simultaneously.
Introduction to multiplexing, packet switching.pptxnamrata110808
Multiplexing refers to combining multiple information signals into one channel for transmission. Demultiplexing is separating the signals at the receiving end. There are four main types of multiplexing: frequency division, wavelength division, time division, and code division. Time division multiplexing transmits signals by assigning each a time slot, while frequency division divides the bandwidth into frequency channels. Packet switching uses statistical multiplexing without a fixed frame structure and may involve queuing delays, making it suitable for bursty traffic.
Multiple access techniques allow multiple users to share finite radio spectrum resources simultaneously. They can be categorized as narrowband or wideband. Common techniques include FDMA, TDMA, CDMA, and SDMA. FDMA divides the total bandwidth into narrow channels that are allocated to users. TDMA divides each channel into time slots that are allocated to users. CDMA spreads the signal over a wide bandwidth using pseudo-random codes and allows multiple signals to overlap in both time and frequency.
This document discusses various techniques for multiplexing in wireless networks. It covers four main types of multiplexing: frequency-division multiplexing (FDM), time-division multiplexing (TDM), code-division multiplexing (CDM), and space-division multiplexing (SDM). For each technique, it provides examples of how they work, their advantages and disadvantages, and applications where they are used including GSM networks which combine FDM and TDM. It also discusses concepts like correlation between codes and how orthogonal codes allow receivers in CDMA networks to distinguish different user signals.
This document discusses multiple-input multiple-output (MIMO) systems, including their motivations and capabilities. MIMO systems use multiple antennas at both the transmitter and receiver to achieve high data rates approaching 1 Gbps while maintaining quality of service. The document covers MIMO channel models and capacity, design criteria like diversity and spatial multiplexing, practical architectures like V-BLAST and Alamouti's scheme, and applications to networking including MIMO-OFDM and MIMO MAC protocols.
Comparitive analysis of bit error rates of multiple input multiple output tra...slinpublishers
The document compares the bit error rates of multiple input multiple output (MIMO) transmission schemes, including spatial multiplexing, space-time block codes (STBC), and space-time block coded spatial modulation (STBC-SM). It finds that STBC-SM provides better performance than STBC and vertical-Bell labs layered space-time (V-BLAST) spatial multiplexing. Specifically, simulations show STBC-SM has a lower bit error rate than the other schemes when using four transmit and four receive antennas. The document explains the techniques of V-BLAST, STBC, and STBC-SM in detail.
Multiple access techniques allow multiple mobile users to simultaneously share a finite amount of radio spectrum for communication. Common techniques include FDMA, TDMA, CDMA, and SDMA. FDMA allocates different frequency bands to different users. TDMA divides the available bandwidth into time slots that are allocated to users. CDMA spreads user signals over the entire available bandwidth through coding.
This document discusses channel estimation techniques in OFDM systems. It compares LS and MMSE estimation methods. It also describes simulating an OFDM transmission in Matlab to analyze how the bit error ratio is affected by changing the signal-to-noise ratio and multipath effects. The key steps of the simulation are outlined, including OFDM transmission, channel estimation using LS and MMSE, and calculating the bit error ratio to compare performance of the estimation techniques.
1. Random access protocols like ALOHA and slotted ALOHA allow wireless stations to transmit randomly, which can cause collisions and reduce throughput. Carrier sensing multiple access (CSMA) protocols reduce collisions by having stations sense the channel before transmitting. (2 sentences)
2. Scheduling-based medium access control protocols like reservation and polling systems organize transmissions to avoid collisions. Reservation systems use minislots for stations to request transmission slots in future frames. Polling systems have a central controller or token that polls stations in a set order to transmit. (3 sentences)
1. Random access protocols like ALOHA and slotted ALOHA allow wireless stations to transmit randomly, which can cause collisions and reduce throughput. Carrier sensing multiple access (CSMA) protocols reduce collisions by having stations sense the channel before transmitting. (2 sentences)
2. Scheduling-based medium access control protocols like reservation and polling systems organize transmissions to avoid collisions. Reservation systems use reservation intervals for stations to request transmission slots, while polling systems have a central controller or passing permit poll stations in a set order. Both approaches can provide more efficient channel utilization than random access. (3 sentences)
How to Manage Your Lost Opportunities in Odoo 17 CRMCeline George
Odoo 17 CRM allows us to track why we lose sales opportunities with "Lost Reasons." This helps analyze our sales process and identify areas for improvement. Here's how to configure lost reasons in Odoo 17 CRM
The simplified electron and muon model, Oscillating Spacetime: The Foundation...RitikBhardwaj56
Discover the Simplified Electron and Muon Model: A New Wave-Based Approach to Understanding Particles delves into a groundbreaking theory that presents electrons and muons as rotating soliton waves within oscillating spacetime. Geared towards students, researchers, and science buffs, this book breaks down complex ideas into simple explanations. It covers topics such as electron waves, temporal dynamics, and the implications of this model on particle physics. With clear illustrations and easy-to-follow explanations, readers will gain a new outlook on the universe's fundamental nature.
Macroeconomics- Movie Location
This will be used as part of your Personal Professional Portfolio once graded.
Objective:
Prepare a presentation or a paper using research, basic comparative analysis, data organization and application of economic information. You will make an informed assessment of an economic climate outside of the United States to accomplish an entertainment industry objective.
A Strategic Approach: GenAI in EducationPeter Windle
Artificial Intelligence (AI) technologies such as Generative AI, Image Generators and Large Language Models have had a dramatic impact on teaching, learning and assessment over the past 18 months. The most immediate threat AI posed was to Academic Integrity with Higher Education Institutes (HEIs) focusing their efforts on combating the use of GenAI in assessment. Guidelines were developed for staff and students, policies put in place too. Innovative educators have forged paths in the use of Generative AI for teaching, learning and assessments leading to pockets of transformation springing up across HEIs, often with little or no top-down guidance, support or direction.
This Gasta posits a strategic approach to integrating AI into HEIs to prepare staff, students and the curriculum for an evolving world and workplace. We will highlight the advantages of working with these technologies beyond the realm of teaching, learning and assessment by considering prompt engineering skills, industry impact, curriculum changes, and the need for staff upskilling. In contrast, not engaging strategically with Generative AI poses risks, including falling behind peers, missed opportunities and failing to ensure our graduates remain employable. The rapid evolution of AI technologies necessitates a proactive and strategic approach if we are to remain relevant.
Exploiting Artificial Intelligence for Empowering Researchers and Faculty, In...Dr. Vinod Kumar Kanvaria
Exploiting Artificial Intelligence for Empowering Researchers and Faculty,
International FDP on Fundamentals of Research in Social Sciences
at Integral University, Lucknow, 06.06.2024
By Dr. Vinod Kumar Kanvaria
A review of the growth of the Israel Genealogy Research Association Database Collection for the last 12 months. Our collection is now passed the 3 million mark and still growing. See which archives have contributed the most. See the different types of records we have, and which years have had records added. You can also see what we have for the future.
1. King Fahd University of
Petroleum & Minerals
Computer Engineering Dept
COE 541 – Design and Analysis of
Local Area Networks
Term 031
Dr. Ashraf S. Hasan Mahmoud
Rm 22-148-3
Ext. 1724
Email: ashraf@ccse.kfupm.edu.sa
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 1
Fixed Assignment Access
• Schemes:
• Time Division Multiple Access (Time)
• Frequency Division Multiple Access (BW)
• Code Division Multiple Access (Code)
• Access to common channel is
independent of user demand – static and
predetermined
• Contrast to asynchronous time multiplexing
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 2
2. TDM – Example2: Digital Carrier
Systems
• Voice call is PCM
coded 8
b/sample
• DS-0: PCM digitized
voice call – R = 64
Kb/s
• Group 24 digitized
voice calls into one
frame as shown in
figure DS-1: 24
DS-0s
• Note channel 1 has
all 1st bits from all
of 24 calls; channel
2 has all 2nd bits
from all 24 calls;
etc.
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 3
TDMA
• Assume:
• M users/stations
• Channel of bit rate = R b/s
• Fixed packet size = X bits/packet
• Packet arrival = λ packet/sec
frame 1 frame 2 frame 3
1 2 3 M 1 2 3 M 1 2 3 M
time
Frame = M slots
= MX/R sec
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 4
3. TDMA – Queueing Model
• Assume: R/M bits/sec
λ
user/station 1
buffer
R/M bits/sec
λ
user/station 2
buffer
R/M bits/sec
λ
user/station M
buffer
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 5
TDMA – Total Delay Analysis
• Total Packet (Burst) delay:
• Slot Synchronization Delay – Avg = ½ frame
duration, plus
• Queueing Delay, plus
• Packet transmission
• Slot Synchronization = X M/(2R)
• Packet transmission = X/R
• Queueing Delay = ?
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 6
4. TDMA – Packet (Burst) Queueing
Delay
• Each channel can be modeled as an
M/D/1 queue
• Consider station/user queues individually
• Poisson arrival of packets of rate λ
• Service time – fixed (packet size is fixed and
so is the transmission rate)
• From point of view of user queue – packet is
transmitted at rate of R/M bits/seconds
• Packet transmission time = X/(R/M) or MX/R
seconds
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 7
TDMA – Packet (Burst) Queueing
Delay – cont’d
• For M/D/1 (refer to M/G/1 slides):
ρ
E[W] = ---------- E[τ]
2 (1-ρ)
E[τ] is the packet service time = MX/R
ρ = λ E[τ] = λ MX/R
• Therefore, the mean waiting time can be written
as
ρ MX
E[W] = ---------- ----
2 (1-ρ) R
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 8
5. TDMA – Throughput
• Throughput: average number of useful
(good) packets transmission per time
unit
• Each station transmits λ X/R packets per
time unit Station throughput = λ X/R
• The M stations community throughput =
M λ X/R
• Total Throughput, S = M λ X/R
=ρ
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 9
TDMA – Total Delay
• Total Delay, T
X MX M S X
T = ---- + ----- + ------------- ----
R 2R 2(1 – S) R
Normalizing total delay with respect to packet
transmission time
M M S
Ť= 1 + ---- + ----------
2 2(1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 10
6. FDMA – Total Delay
• Assume same traffic parameters (for
comparison reasons)
• No slot synchronization time –
transmission can be always on
• Total Packet (Burst) delay:
• Queueing Delay, plus
• Packet transmission
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 11
FDMA – Total Delay – cont’d
• Total delay, T
MX S MX
T = ----- + ------------- -----
R 2(1 – S) R
• Or
M(2 – S)
Ť = -------------
2(1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 12
7. TDMA versus FDMA – Total Delay
• Using the previous relations,
ŤFDMA = ŤTDMA + M/2 – 1
• i.e. total delay for FDMA is always greater
than that for TDMA except for M = 2
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 13
Performance Measures
• Throughput
• Delay (packet)
Throughput
Ideal Load-Throughput Relation
C
C Offered Load
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 14
8. Pure ALOHA
Central node
station
nodes
station
station
station New packets S
G S
channel collision
old packets N
Uplink carrier 413 kHz, 9.6 kb/s
Downlink carrier 407 kHz, 9.6 kb/s Y
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 15
ALOHA Random Access Procedure
• Assume
• Packet transmission time: P
• Total # of packet arrival (new +
retransmission) ~ Poisson with rate λ
1 2 M
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 16
9. ALOHA - Throughput
• Poisson arrival (new + retransmitted) of
packets:
(λt)k
Prob[k arrivals in t sec] = ------ e –λt
k!
• Offered Load (G): Average number of attempted
packet transmissions per packet transmission
time, P
• Throughput (S): Average number of successful
transmissions per packet transmission time, P
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 17
Pure ALOHA – Throughput – cont’d
• Vulnerable Period
User of interest
t1 t1 + P time
Other users
t1 - P t1 t1+ P time
Any transmission in this
period will lead to collision
Vulnerable Period = 2 P
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 18
10. ALOHA Throughput – cont’d
• Throughput = fraction of attempted transmission that are
successful (i.e. did not collide)
• Therefore,
S = G X Prob[ no collision in 2 P seconds]
= G X Prob[0 packet arrivals in 2 P seconds]
(2G)0
= G X ----- e -2G
0!
Or
S = G e -2G packets/packet transmission time
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 19
Slotted ALOHA
• An improvement over pure ALOHA
• Time axis is slotted
• Transmission occur only at the beginning
of a time slot
• A packet arriving to buffer has to wait till
the beginning of the time slot for
transmission
• Cost: common clock signal!
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 20
11. Slotted ALOHA – Throughput –
cont’d
• Vulnerable Period (note time axis is divided into slots –
transmissions can only start at the beginning of a time
slot)
User of interest
t1 t1 + P time
Other users
t1 - P t1 t1+ P time
No arrivals should
occur during this slot
Any transmission at this
Vulnerable Period = 1 P
12/21/2003 Dr. Ashraf S. Hasancollision
instant will lead to Mahmoud 21
Slotted ALOHA – Throughput
• Throughput = fraction of attempted transmission that are
successful (i.e. did not collide)
• Therefore,
S = G X Prob[ no collision in 1 P seconds]
= G X Prob[0 packet arrivals in 1 P seconds]
(G)0
= G X ----- e -1G
0!
Or
S = G e -1G packets/packet transmission time
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 22
12. ALOHA – Throughput – cont’d
• Pure ALOHA: Max throughput, S = 0.5 e-1 or ~ 18% at G
=½
• Slotted ALOHA: Max throughput, S = e-1 or ~ 36% at G =
1
Pure ALOHA / Slotted ALOHA Throughput Curves
0.4
Pure ALOHA
•For Pure ALOHA:
Slotted ALOHA
0.35
Successful packet transmission / time unit
• Stable operation 0.3
range: 0 < G < 0.5 0.25
• Unstable operation 0.2
channel saturation
range: G > 0.5
0.15
•For Slotted ALOHA: 0.1
• Stable operation 0.05
range: 0 < G < 1.0
0
• Unstable operation 0 0.5 1 1.5 2 2.5 3 3.5 4
Load - Attempted packet transmissions / time unit
4.5 5
range: G > 1.0
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 23
Pure ALOHA – (Approximate) Delay
Analysis
• Average number of attempts per
successfully transmitted packet = G/S
• From throughput relation,
G/S = e2G
• Therefore, average number unsuccessful
attempts = G/S – 1
= e2G – 1
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 24
13. Pure ALOHA – (Approximate) Delay
Analysis – cont’d
• Cost for each collision
• Backoff time - assume duration B on average
• Retransmission
• Therefore, total delay, T
T = P + (e2G - 1)(P + B)
• Normalizing the total delay yields,
Ť = 1 + (e2G - 1)(1 + B/P)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 25
Example 1
• Problem: A centralized network providing a
maximum of 10 Mbps and services a large set
of user terminal with pure ALOHA protocol
a) What is the maximum throughput for network?
b) What is the offered traffic in the medium and
how is it composed?
c) If a packet length is 64KBytes, what is the
average packet delay? Assume average backoff
time = 1 second.
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 26
14. Example 1 – cont’d
• Solution:
a) Smax = 18%
==> Network throughput = 0.18 X 10 = 1.8 Mbps
b) At S = Smax, G = 0.5,
Offered load = 0.5 X 10 = 5 Mbps
Composition of load: 1.8 Mbps of delivered packets
+ 3.2 Mbps of collided packets
c) Packet transmission time P = 64X1024X8 bits/10 Mb/s
= 6.6 msec
T = P + (e2G - 1)(P + B)
= 6.6 + (e1 – 1)(6.6 + 1000)
= 1736
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 27
Notes On ALOHA Analysis
• Slotted ALOHA: a modified ALOHA
protocol to allow stations to transmit
only at known and fixed time instances.
• Time axis is divided into slots – stations can
transmit only at the beginning of a time slot
• What is the vulnerable period for slotted
ALOHA?
• Derive the throughput and delay
relationship for this protocol?
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 28
15. Idealized Central Control
• Idealized = ZERO cost for transfer of
channel from one state to another under
central node
• Whenever a station has data to transmit,
controller knows instantly and the channel
assignment is immediate
• Packets arriving while channel is busy are
queued (infinite buffer)
• If two stations have queues packets, the
one with first arrival is chosen
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 29
Idealized Central Control - Analysis
• Assumptions (same as before):
• Arrival at each station ~ Poisson of λ packets/sec
• Packets have constant length of X bits
• M stations
• Channel bit rate = R b/s
• Propagation and processing times ≈ 0
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 30
16. Idealized Central Control – Analysis
– cont’d
• Total input = M λ packets / second
• Since “no cost” for transfer of channel ==>
distributed network behaves like a single queue
• Over all throughput (utilization) is given by
ρ = M λ * (X/R)
where M λ is the total arrival rate to this single
queue, and
X/R is the service time
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 31
Idealized Central Control – Analysis
– cont’d
• This single queue – M/D/1
• Mean number of queued packets (E[Nq]) and
mean waiting time (E[W])
ρ2 ρ X
Nq = ---------, W = --------- ----
2(1 – ρ) 2(1 – ρ) R
• Therefore, total delay, T, is given by
X ρ X
T = ----- + ---------- -----
R 2(1 – ρ) R
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 32
17. Idealized Central Control – Analysis
– cont’d
• Since there are no collisions ==> throughput = utilization or S =
ρ
• Hence, total delay is given by
X S X
T = ----- + ---------- -----
R 2(1 – S) R
Or
S
Ť = 1 + ----------
2(1 – S)
(2 – S)
= -----------
2(1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 33
Idealized Central Control – Analysis
– cont’d
• Also, E[Nq] is given by
S2
E[Nq] = ----------
2(1 – S)
• Per station throughput = S/M
• Per station nq = E[Nq] / M
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 34
18. Polling Networks
• Central Control Networks: a central node
arbitrates access to the network
• The access order is predetermined –
under the control of the central node
• Access is granted when station is polled –
Full rate of channel is used
• Stations accumulate traffic in their buffers
• Transmit when given permission (polled)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 35
Operation Modes
• Two Modes:
1. Roll-Call
2. Hub polling
• For the two modes, the opportunity to
transmit is symmetrically rotated from
one station to another
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 36
19. Operation Modes – Roll-Call
• Central node initiates polling sequence
by sending polling message to chosen
station
• Polled station transmits traffic (if any)
• Transmitting station informs central node
of transmission end (field in the last
transmitted packet)
• Central node polls next station in-line
• Process repeats
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 37
Operation Modes – Hub polling
• Central node initiates polling sequence
by sending polling message to chosen
station
• Polled station transmits traffic (if any)
• Last transmitted packet contains a go-
ahead signal with the next inline station
address
• Next inline station (which is continuously
monitoring traffic) identifies its address
and starts transmitting (if there is traffic
in buffer) immediately
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 38
20. Roll-Call vs Hub based
• Response time
• Complexity – Cost
• ?
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 39
Logical Structure
Station i
direction
of poll
Station 2
Station M
Central
Station 1 node
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 40
21. Performance Analysis
• Assume:
• Arrival process is Poisson with rate λ
packets/sec
• The walk time, w, between station stations
is constant
• Includes processing and propagation time
• Average packet length = Xavg bits/packet
• Will consider fixed and exponentially distributed
packet sizes
• Common channel (server) rate = R b/s
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 41
Performance Analysis – cont’d
• Cycle Time, Tc:
• Total time to poll each station and return to
the starting station in the polling sequence
• Random variable
• Amount of data transmitted by each station is
random
• Other performance measures:
• Average queue length, N, in station (packets)
• Average time, W, that packets wait in the
station buffer before being transmitted
• Average transfer delay, T, from packet entry
into station buffer till delivery to central node
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 42
22. Cycle Time
• Let Nm be the average number of packets
stored in station buffer
• Nm includes packets arriving to buffer while
station is in service
• Time to empty buffer = Nm Xavg /R
• Cycle Time, Tc
Tc = M [ Nm Xavg / R + w ]
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 43
Cycle Time – cont’d
• At steady state, Nm is given by
Nm = λ Tc
• Substituting in the previous equation yields
Mw
Tc = --------------------
1 - M λ Xavg / R
Or
Mw
Tc = ------------- seconds
1-S
where
throughput S = (M λ) / (R/Xavg) < 1 or (M λ) < (R/Xavg)!!
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 44
23. Delay Analysis
• Packet waiting time, W, in queue:
• Waiting time in queue, W1, while other
stations are being served, plus
• Waiting time in queue, W2, while its station is
being served and till packet reaches head of
queue
packet arrival
service station i service station i
W1 time
W2
W
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 45
Delay Analysis – cont’d
• Avg number of packets transmitted by station in a
cycle: Nm = λ Tc
• remember we serve till buffer is empty
• Average service time for station equals to
λ Tc Xavg/R
• Define ρ as
ρ = λ Xavg / R
• Therefore, average service time per station is
given by
ρ Tc
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 46
24. Cycle for a Polling Network
• Note the cycle time Tc partitioning
w w
service service service service service
station i station i+1 station M station 1 station i
(1-ρ)Tc ρTc time
Tc
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 47
Delay Analysis – cont’d
• (1-ρ)Tc is the average time station I waits
to be served
• Packet arrive at random during (1-ρ)Tc
• Average waiting time W1 = (1-ρ)Tc/2
• Substitute the expression for Tc, yields
Mw(1 – ρ)
W1 = --------------
2(1 – M ρ) It remains to compute W2!
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 48
25. Delay Analysis – cont’d
• Writing W1 in terms of S = M λ Xavg/R
Mw(1 – S/M)
W1 = -----------------
2(1 – S)
It remains to compute W2!
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 49
Delay Analysis – cont’d
• To determine W2 – consider the following
equivalent queueing system
• Server never goes idle (no walk time) – switches
instantly from one buffer to the next
• All arrivals aggregated
• All buffers lumped
• This model: M/G/1
Mλ
ALL network
arrivals
ALL buffers lumped τ = Xavg / R packets/sec
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 50
26. Delay Analysis – cont’d
• For an M/G/1 with arrival rate λ and service time, τ:
average waiting time, E[W], is given by
λE[τ2]
E[W] = ----------
2(1-ρ)
• For our hypothetical queue:
• λ Mλ
• E[τ] = Xavg/R; E[τ2] = E[X2]/R2
• Therefore, W2 is given by
M λ E[X2]/R2
W2 = ----------------
2 ( 1 – Mρ )
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 51
Delay Analysis – cont’d
• Writing W2 in terms of S = M λ Xavg/R
S E[X2]
W2 = ----------------------
2 Xavg R ( 1 – S )
Therefore, overall waiting time for the packet:
W = W1 + W2
Mw(1 – S/M) S E[X2]
W = ----------------- + ------------------
2(1 – S) 2E[X] R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 52
27. Delay Analysis – Constant Packet
Size
• For constant packet size X
• E[X] = X
• E[X2] = X2
• Therefore, overall waiting time for the packet:
Mw(1 – S/M) S E[X]
W = ----------------- + --------------
2(1 – S) 2R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 53
Delay Analysis – Exponential
Packet Size
• For exponentially distributed packet sizes, X
• E[X] = Xavg
• E[X2] = 2 (Xavg) 2 = 2 E[X]2
• Therefore, overall waiting time for the packet:
Mw(1 – S/M) S E[X]
W = ----------------- + --------------
2(1 – S) R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 54
28. Example 2:
• Problem: Consider a metropolitan area network with a single
central processor located at the headend of a broadband CATV
system that has a tree topology. The following are specified:
• Maximum distance from headend to subscriber station = 20 km
• Access technique – roll-call polling
• Length of polling packet = 8 Bytes
• Length of go-ahead packet = 1 Bytes
• Data rate of channel = 56 kb/s
• Number of subscribers = 1000
• Packet length distribution for packets from subs to headend –
exponential
• Mean packet length = 200 Bytes
• Propagation delay = 6 µsec/km
• Modern sync time = 10 msec
A. Find the mean waiting delay for arriving packets at the stations if
each user generates an average of one packet per minute
B. If the channel rate is reduced to 9600 b/s what is the longest
possible mean packet length that will not overload the system?
C. For mean packet lengths of two-thirds the result of (B)
determine the mean waiting delay
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 55
Example 2: cont’d
• Solution:
Mean walking time, w:
w = transmission time of go-ahead packet* +
propagation delay from station to headend +
transmission of polling packet +
propagation delay from headend to next station +
modern sync time
One way propagation = 20 X 6 = 120 µsec
Transmission time for go-ahead packet = 1 X 8 /56 = 0.14
msec
Transmission time for polling packet = 8 X 8 / 56 = 1.14
msec
Therefore: w = 0.14 + 2 X 0.120 + 1.14 +10 = 11.52 msec
*This decomposition of the walk time assumes there is a separate go-ahead packet indicating end of traffic condition
– alternatively, the last traffic packet could convey the same information by setting a flag
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 56
29. Example 2: cont’d
• Solution:
A) Mean waiting delay, W is given by
Mw(1 – S/M) S E[X]
W = ----------------- + --------------
2(1 – S) R (1 – S)
We need to compute S first –
S = M λ Xavg/R
= 1000 X (1/60) X 200 X 8 /56 = 0.476
Substituting in the formula for W, yields
W = 10.99 + 0.026
= 11.02 seconds
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 57
Example 2: cont’d
• Solution:
B) Smax <= 1 M λ Xavgmax/R <= 1
For R = 9600 b/s
Xavgmax <= R/(M λ) = 72 Bytes
C) For Xavg = 2/3 Xavgmax
= 2/3 (72) = 48 Bytes
S = M λ Xavg/R = 0.667
The new walking time, w is given by
w = 8X8/9.6 + 1X8/9.6 + 2X0.12 +10 = 17.74 msec
Use the new values for S and w and sub in the expression for W
W = 26.62 + 0.01 = 26.63 seconds
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 58
30. Average Number of Packets Per
Station
• Using Little’s formula:
M λ w(1 – S/M) S λ E[X2]
N = ----------------- + ------------------
2(1 – S) 2E[X] R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 59
Average Number of Packets Per
Station – Constant Packet Size
• For constant packet size X
• E[X] = X
• E[X2] = X2
• Therefore, overall waiting time for the packet:
M λ w(1 – S/M) S λE[X]
N = ------------------- + --------------
2(1 – S) 2R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 60
31. Average Number of Packets Per
Station – Exponential Packet Size
• For exponentially distributed packet sizes, X
• E[X] = Xavg
• E[X2] = 2 (Xavg) 2 = 2 E[X]2
• Therefore, overall waiting time for the packet:
M λ w(1 – S/M) S λE[X]
N = -------------------- + --------------
2(1 – S) R (1 – S)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 61
Example 3:
• Problem: For the network specified in Example 2, find the average
number of packets per station for parts (A) and (C).
• Solution:
(A) w = 11.52 msec, S = 0.476, M = 1000 – exponential packet sizes
1000X 0.01152/60(1-0.476/1000) 0.476/60X200X8
N = ---------------------------------------- + --------------------
2 ( 1 – 0.476 ) 56000 X(1 – 0.476)
= 0.183 + 0.000432
= 0.183 packets / station
(C) w = 17.74 msec, S = 0.667, M = 1000 – exponential packet sizes
N = 0.444 + 0.00133
= 0.445 packets / station
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 62
32. Adaptive Polling
• Using waiting time and buffer size
equations: under light to moderate
loading (i.e. S is small) – performance
depends mainly on Number of stations,
M and walking time, w
W ~ Mw/2
• Try to reduce number of polls
Adaptive cycles
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 63
Adaptive Polling: Pure Probing
• Nodes are organized in a tree structure
1
• Controller carries out
probing procedure by
Example: M = 8 2
separating stations
into 2 groups that are
probed at one at a
# of probes required =7 3
time by a signal
broadcast to all 2nd probe
st 4
stations in that group 1 probe
• If a +ve response is
received from a 5
3rd probe 6th probe
group, it is further
divided into 2
4th probe 6 7th probe
subgroups
• Process continues till Station with message to send
station is identified 5th probe 7
Station with no message to send
12/21/2003 Dr. Ashraf S. Hasan Mahmoud
864
33. Adaptive Polling: Pure Probing –
cont’d
• Designed for low load conditions – i.e. a small
fraction of terminals are transmitting
• Controller does not know that only one station
wants to transmit
• If the number of stations = M
• 2Xlog2(M) + 1 probes are needed to locate a
single ready user
• Remember a standard polling requires M = 2n
polls at most (M/2 = 2n-1 on average to locate the
single ready user)
• Example: M = 256 stations:
• Pure probing: 17 probes
• Standard polling: 256 polls
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 65
Adaptive Polling: Pure Probing –
cont’d
• When more than one station has data –
number of probes increase
• Under heavy load (i.e. all stations have
data to transmit) – number of probes
becomes equal or greater than number
of polls for standard polling
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 66
34. Ring Networks
• Based on network geometry
• Characterized as a sequence of point-
to-point links between stations, closed
on itself.
• All messages travel over a fixed route
from station to station around the loop
• Interface unit connects station to ring
• Regenerate messages and identifies
addresses
• Does not store messages
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 67
Ring Networks
• Station latency – few bit times for all
traffic passing through message
(processing time)
• Typically – ring = high
speed directional bus 1
• Propagation delay ~ 0 Ring interface
unit
4 2
station
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 3 68
35. Ring Networks - Advantages
• Simple implementation
• No routing is required
• Only a small latency added
• Can cover large distances (metropolitan
area networks) – signal/message
regeneration
• Efficiency does not degrade rapidly with
load
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 69
Ring Networks - Disadvantages
• Single point failure – if a single station
interface fails …
• Not so easy to expand/modify – ring
must be broken
• Propagation delay is proportional to
number of stations
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 70
36. Types of Ring Networks
• Three Basic Types:
1. Token Rings: control access to ring
through passing of ring from station to
station – almost same as hub polling
2. Slotted Rings: a small number of fixed-
sized slots are circulated; when empty
they are available for use by any station
3. Register Insertion Rings: two shift
registers for each station node as switches
to control traffic into and out of the ring –
long packets can be served
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 71
Token Ring Networks
• Access to ring is controlled by a token
• Token states: busy or idle
• When ring is first activate – a master
station circulates an idle token
• To transmit data, a station must:
• Capture token
• Set token to busy
• Transmit data
• Set token to idle
See http://www.datacottage.com/nch/troperation.htm for a basic introduction
and animated operation or better check http://www.macs.hw.ac.uk/~pjbk/commbook/lans/
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 72
37. Token Ring Networks
• Same basic structure for all rings
ring output line line ring input
controller
driver receiver
1 Transmitter Receiver
Note: textbook has
typos in this figure
Ring interface transmit Receive
unit buffer buffer
Attached Device
4 2
station
listen mode transmit mode
3 delay delay
ring ring ring ring
output input output input
node out node in node out node in
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 73
Token Pattern
• Token:
• A dedicated pattern of several bits, or
• A single bit transmitted in a format different that
that used for data bits
• Example: IEEE802 – token = several bytes
long
• Bit stuffing is used to prevent occurrence of
similar patterns
• Usually, one bit in this pattern is used to
indicate whether the token is busy or free
• To set the token bit – station latency = 1 bit time
• Can be used to add priority functionality
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 74
38. Service Discipline
• Exhaustive
• Station retains use of ring until it has transmitted all
the data stored in transmit buffer
• Non-exhaustive
• Station is allowed to transmit only a specified
number of bits each time it captures the token
• Two disciples provide same performance for
light-medium loads
• The analysis in this package assumes
exhaustive
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 75
Idle Operation
• Synchronization
• Use of Manchester encoding
• All stations in listen mode
• Token circulates around the ring
• Ring Latency = propagation time + sum of
station latencies
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 76
39. Normal Operation
• One station captures token
• Station transmits data
• Station produces a modified token (or a
control field in the header of the data
packet) to indicate to other stations that
ring is not free (i.e. token is part of
packet)
• Transmitting station is responsible for:
• removing its packet from the ring, and
• generating a new packet
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 77
Normal Operation – cont’d
• When the new token is generated –
leads to three different modes of
operation
• Multiple token,
• Single token, and
• Single packet operation
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 78
40. Multiple Token Operation
• The transmitting station generates a
new FREE token and places it on the ring
immediately following the last bit of
transmitted data
• This permits several busy tokens on the
ring!!
• What are the packet times in relation to
ring latency required to achieve this?
• But only one free token exists!!
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 79
Single Token Operation
• The transmitting station generates a new FREE
token and places it on the ring immediately
ONLY after it removes its BUSY token
• Two Cases arise:
• Packet time > ring latency: station will receive (and
erase) its busy token before it has finished
transmitting its packet – new FREE token generated
after packet is completed – looks the same as
multiple token operation
• Packet time < ring latency: station will finish
transmission of packet – must wait till it receives
(and erase) busy token – new FREE is then
generated
• Only a single token exits on the ring at any
time
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 80
41. Single Packet Operation
• The transmitting station does not issue a
new FREE token until after it has
circulated completely around the ring
and erased all of its transmitted packet
• Same as single token operation except here
also the packet has to be removed before
the new token is generated
• Only a single token exits on the ring at
any time
• Very conservative behavior – no two
simultaneous transmissions on the ring
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 81
Example 4: Four-Station Token Ring
• Example:
1
In #1 Out #1
Out #4
4 2
Out #2
Out #3
3
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 82
42. Ring Networks – Token Ring
Multiple Token Single Token Single Packet time
time
In #1 Out #1 Out #2 Out #3 Out #4 In #1 Out #1 Out #2 Out #3 Out #4 In #1 Out #1 Out #2 Out #3 Out #4
0 0
2 2
d d d
4 d d d d d d 4
d d d d d d d d d
6 d d d d d d d d d d d d d d d 6
d d d d d d d d d d d d d d d
8 d d d d d d d d d d d d 8
d d d d d d d d d
10 d d d d d d 10
12 d d d d 12
d d d d d d
14 d d d d 14
d d d d d d
16 d d d d d 16
d d
18 d d 18
d
20
free token
22
busy token
24
12/21/2003 d data bit Dr. Ashraf S. Hasan Mahmoud 83
26
Token Ring - Delay Analysis
• Assumptions
• All stations are identical load-wise
• Arrival process ~ Poisson with λ packets /
second / station
• There are M stations
• The average distance between stations ≈
one-half the distance around the ring
• Propagation delay between consecutive
stations = τ/M – where τ is the total ring
propagation time
• Packet size: random (uniform or exp) –
average packet size = Xavg bits / packet
• Exhaustive service time
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 84
43. Token Ring - Delay Analysis
• Assumptions – cont’d
• Channel bit rate, R bits / second
• Latency per station B bits
• Round trip propagation = τ seconds
• Ring Latency = τ’
• Required: Determine the transfer delay
for token passing rings (multiple token,
single token, and single packet)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 85
Token Rings vs Hub polling
• Difference:
• Token ring has no central
station/controller
• Similarities:
• Walk time in hub polling equivalent to time
from packet transmission finish till instant
when next station receives free token
• Therefore we will adapt the hub polling
performance equations to our case here
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 86
44. Review: Hub polling Performance
• It was shown previously, the packet waiting time for a
polling network is given by
Mw(1 – S/M) S E[X2]
W = ----------------- + ------------------
2(1 – S) 2E[X] R (1 – S)
Where S – is the network throughput
• The average Transfer delay (i.e. Waiting plus service
time) is given by
T = Xavg/R + τavg + W
Where τavg is the average propagation delay from station to
the central computer in the polling network
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 87
Token Ring Performance
• Ring Latency:
τ’ = total propagation time + sum station
latencies (refer to slide 76)
τ’ = τ + M B/R
• One average a transmission will face τ’/2 of
latency before being received
• Therefore, for token ring, transfer delay T is
given by
T = Xavg/R + τ’/2 + W
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 88
45. Token Ring Performance – cont’d 2
• To compute W for token ring, we need to find:
• The equivalent walk time
• The network throughput
• The moments for service time: E[X]/R, and
E[X2]/R2
• Walk time,
w = propagation delay from station to the
next + station latency
= τ/M + B/R
= τ’/M
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 89
Token Ring Performance – cont’d 3
• Define “effective throughput”, S’ to be
S’ = Mλ E[EST]
(remember throughput for the network is defined as
S = Mλ Xavg/R )
where E[EST] is the average effective service time
for a terminal on the ring
E[EST] = total time consumed by the ring to
process one packet and become free to
process the next packet
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 90
46. Token Ring Performance – cont’d 4
• Therefore, total transfer delay, T is
given by
T = Xavg/R + τ’/2 + W
OUR MAIN RESULTS for
and W is given by RING NETWORKS
τ’(1 – S’/M) S’ E[EST2]
W = ----------------- + ------------------
2(1 – S’) 2E[EST] (1 – S’)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 91
Token Ring Performance – Multiple
Token Operation
• For multiple token operation, a free token is generated
immediately after the last data bit is transmitted
E[ESTmultiple_token] = ESTavg = Xavg/R
E[ESTmultiple_token2] = E[X2]/R2
• Therefore, the total transfer delay, T is given by
Tmultiple_token = Xavg/R + τ’/2 + Wmultiple_token
where
τ’(1 – S/M) S E[X2]
Wmultiple_token = ----------------- + ------------------
2(1 – S) 2E[X] R (1 – S)
12/21/2003 We all know by nowDr. Ashraf S. Hasan Mahmoud
how to evaluate E[X2] and E[X] for constant/uniform/ 92
exponentially distributed packet sizes – refer to slides 53-54.
47. Token Ring Performance – Single
Token Operation
• For single token operation, a free token is
generated when the busy token has circulated
the ring completely!
• To evaluate E[ESTsignle_token] let us define the
normalized ring latency parameter a’
a’ = τ’ / (Xavg / R)
= propagation time/ (Xavg/R) +
(M B /R) / (Xavg / R)
= a + M B / Xavg
a is the normalized ring propagation time (i.e.
Tprop / Tframe)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 93
Token Ring Performance – Single
Token Operation – cont’d
• The two cases that arise:
• a’ < 1 busy token will be received before packet
transmission is completed
• a’ > 1 packet transmission time finishes before
start of packet circulates the ring
• This is related to the packet size X
• X can be constant
• X can be exponentially distributed
• Each of these cases will be considered
separately
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 94
48. Token Ring Performance – Single
Token Operation – Constant Packet
Size and a’ > 1
• Packet size = X = constant Xavg = X
• Single token operation is the same as
multiple token operation
• Transfer delay, Tsingle_token is the same as
that for Tmultiple_token
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 95
Token Ring Performance – Single
Token Operation – Constant Packet
Size and a’ < 1
• Packet size = X = constant Xavg = X
• Single token operation is different than
the operation of multiple token
• Effective Service Time (EST) = τ’
which is the time for the busy token to
circulate the ring
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 96
49. Token Ring Performance – Single
Token Operation – Constant Packet
Size and a’ < 1 – cont’d
• Therefore,
S’ = Mλ E[EST]
= Mλ τ’ Maximum achievable
= Mλ (Xavg/R) a’ throughput =1 if a’ < 1
1/a’ if a’ > 1
= S a’
• Hence, transfer delay, Tsingle_token, is given by
Tsingle_token = Xavg/R + τ’/2 + Wsingle_token
where
τ’(1 – Sa’/M) Sa’τ’
Wsingle_token = ----------------- + --------------
2(1 – Sa’) 2 (1 – Sa’)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 97
Token Ring Performance – Single
Token Operation – Exponential
Packet Size
• The packet size is random with
exponential distribution
• i.e. For some packets a’ > 1, and for others
a’ < 1
• Therefore, we will use the pdf of the
packet size to find the pdf (or cdf) of EST
and then the E[EST] and E[EST2]
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 98
50. Token Ring Performance – Single
Token Operation – Exponential
Packet Size – cont’d
• Packet size X is
exponentially distributed
0 x<0
f X ( x) = 1 PDF for X
X exp(− x / X ) x ≥ 0
Or
0 x<0
FX ( x) = CDF for X
1 − exp(− x / X ) x ≥ 0 i.e. Prob[ X ≤ x ]
Where E[X] = Xavg = X
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 99
Token Ring Performance – Single
Token Operation – Exponential
Packet Size – cont’d 2
• Service time = X/R
0 x<0
f X / R ( x) = R × f X ( Rx) = R PDF for X/R
( )
X exp − Rx / X x ≥ 0
Or
0 x<0
FX / R ( x) = CDF for X/R
1 − exp(− Rx / X ) x ≥ 0 i.e. Prob[ X/R ≤ x ]
Where E[X/R] = Xavg/R = X / R
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 100
51. Token Ring Performance – Single
Token Operation – Exponential
Packet Size – cont’d 3
• Effective Service time, EST
0 x <τ '
1 − exp(− Rτ ' / X ) x = τ '
PDF for EST
f EST ( x) =
Or R exp(− Rx / X ) x > τ '
X
0 x <τ'
FEST ( x) = CDF for EST
1 − exp(− Rx / X ) x ≥ τ ' i.e. Prob[ EST ≤ x ]
∞
X
E[ EST ] = ∫ x f EST ( x) = exp(− a') + τ '
0
R
∞ 2
X
E[ EST ] = ∫ x f EST ( x) = 2 (1 + a' ) exp(− a') + (τ ')
2 2 2
R
12/21/20030 Dr. Ashraf S. Hasan Mahmoud 101
Token Ring Performance – Single
Token Operation – Exponential
Packet Size – cont’d 4
• Hence, transfer delay, Tsingle_token, is given by
Tsingle_token = Xavg/R + τ’/2 + Wsingle_token
Maximum achievable
where throughput = 1/(e-a’+a’)
τ’[1 – S(e-a’ + a’)/M]
Wsingle_token = ------------------------- +
2[1 – S(e-a’ + a’)]
Xavg S[(a’)2 + 2(1+a’)e-a’]
------ -------------------------
R 2 [1 – S(e-a’ + a’)]
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 102
52. Token Ring Performance – Single
Packet Operation
• For single packet operation, a free token is not
generated until the sending station has
received and erased all of the packet it has
transmitted
• Therefore, ESTsignle_packet is always equal to X/R
+ τ’
• Hence,
E[ESTsignle_packet] = Xavg / R + τ’
E[ESTsignle_packet2] = E[(X/R)2] +2 τ’ E[X]/R + (τ’ )2
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 103
Token Ring Performance – Single
Packet Operation – cont’d
• Hence, transfer delay, Tsingle_packet, is given by
Tsingle_packet = Xavg/R + τ’/2 + Wsingle_packet
Maximum achievable
where throughput = 1/(1+a’)
τ’[1 – (1 + a’)S/M]
Wsingle_packet = ------------------------- +
2[1 – (1 + a’)S]
Xavg S (1+a’) 2
------ ----------------------
R 2 [1 – (1 + a’)S]
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 104
53. Token Ring Performance –
Summary
Ring τ = total round trip propagation time M = number of stations
Parameters: (seconds) B = token size (bits)
τ’ (ring latency) = τ + MB/R (seconds) R = channel bit rate (b/s)
w (equivalent walk time) = τ’/M EST – effective service time
a’ (normalized ring latency ) = τ’/(Xavg/R)
Performance: T = Xavg / R + τ’/2 + W
t’ (1-S’/M) S’ E[EST2]
W = --------------- + --------------------
2(1-S’) 2 E[EST] (1-S’)
Multiple Tokens EST = X/R E[EST] = Xavg/R; E[ESR2] = E[X2]/R2
S’ S
Single Token – If X/R > τ’ same as multiple tokens
Constant X If X/R < τ’ EST = τ’, E[EST] = τ’ and E[EST2] = τ’2
S’ Sa’
Single Token – EST = τ’ if X/R < τ’
Exponential X X/R if X/R > τ’
E[EST] = (Xavg/R) e–a’ + τ’ , E[EST2] = (τ’)2 + 2(Xavg/R)2 e-a’ (1+a’)
S’ S(e-a’ + a’)
Single Packet EST = X/R + τ’ E[EST] = (Xavg/R) + τ’ , E[EST2] = (τ’)2 + 2τ’(Xavg/R) + E[X2]/R2
S’ S(1+a’)
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 105
Example 5:
Problem: For both constant and exponential packets,
evaluate the mean transfer delay for a single-token ring,
that has the following parameters:
• Ring length of 1 km
• Bit rate of 4 Mb/s
• Mean packet length of 1000 bits
• M = 40 stations
• Poisson arrival process to each station with 10
packets/second arrival rate; and
• Station latency of 1 bit
Repeat this calculation for a ring in which the latency is 10
bits.
If the number of stations on the ring is increased from 40 to
120 with the same ring length, evaluate the mean
transfer delay for cases of 1- and 10-bit station latency;
All other network parameters are unchanged
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 106
55. Slotted Rings
• Bits are transferred in serial fashion in one
direction from one station to station around the
ring
• Constant number of bit positions grouped into
fixed-lengths slots – circulate continuously
around the ring
• i.e ring latency measure in bits ≥ total number of bit
positions circulating the ring
• Bit spaces are grouped into mini packets
• Each minipacket contains a bit in the header – bit = 1
occupied; bit = 0 free
• If the slot is empty, it is available for use by a
station with data to transmit
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 109
Slotted Rings – Example 6
• Assume:
• Ring speed (R) = 10 Mb/s (or bit time = 0.1 µsec)
• M = 50 stations
• B = 1 bit
• 2 km ring
• Propagation delay = 5 µsec / km total round
trip 10 µsec
• Ring latency (t’) = 10 + MB/R = 15 µsec
= 150 bit times
Therefore, ring can support: 3 X 50 bit slots, or
4 X 35 bit slots (with
10 bit gap), or
etc.
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 110
56. Slotted Rings – Characteristics
• Designed to transmit relatively few bits
at a time from each station!!
• Minimum access delay
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 111
Cambridge Slotted
• Ring sections coupled
with repeaters
ho
st
• Data rate ~ 10 Mb/s
st
ho
bo cess
• Voice grade twisted pairs
ac box
ac
ss
x
ce
cable – max section
un tion
s ta
tio t
sta uni
n
length = 100 meters
it
R
• Can use coaxial or fiber
R
R
• Monitor station – setup
and maintain ring framing
R
R
– ring manager
• Station unit –
R
R
R
sta uni
un tion
independent transmit and
tio t
it
sta
n
receive modules
ac box
bo cess
monitor
ce
• Access box – interface
x
ac
ss
logic to host
ho
st
ho
st
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 112
57. Cambridge Slotted – cont’d
• Receive module
• Continuously reading signal from repeater
• When a minipacket is addressed to station,
minipacket is saved in receive register
• Minipacket maybe marked to indicate “station is
busy” if station did not copy into receive register –
i.e. was busy
• Transmit module
• Shift register in station unit coupled in parallel to the
access box
• Data and destination bytes are written in parallel to
register – source & control bits added automatically
• A signal from access box sends the content of
register onto the ring to fill the first empty slot
• Transmit register retains a copy of the transmitted
minipacket
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 113
Cambridge Slotted – Minipacket
Format
1 F M Destination Source Data Data R R P
• Total length = 38 bits – 16 bits of data
• Four slots + a short gap (several digits)
• Frame circuit in station – synchronizes with the gap and leading
1 of each minipacket
• Destination – 1 byte
• Source – 1 byte
• Data – 1 byte (for each data field)
• M – monitor
• F – Full/empty bit
• R – Response bits (dest absent, packet accepted, dest deaf, or
dest busy) – read by transmitting station before it decided to
discard its copy of minipacket – no need to ANK/NAK packets
• P – Parity bit
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 114
58. Fairness Requirement
• The full/empty indicator must be
changed to empty after the minipacket
has made a complete circulation of the
ring
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 115
Slotted Ring Operation Example 7
• Two conditions:
• One station has large data packet to transmit
• Two stations have large data packet to transmit
• M=4
• B = 1 – 1 bit station latency
• Propagation delay is ignored
• One slot on the ring
• 1st bit of the 4 bit slot is used the full/empty
indicator
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 116
59. Slotted Ring Operation Example 7
Only #1 is transmitting #1 and #3 transmitting
1
time In #1 Out #1
In #1 Out #1 Out #2 Out #3 Out #4 In #1 Out #1 Out #2 Out #3 Out #4
Empty
0 Out #4
slot
4 2
2 d1 d1
d1 d1 d1 d1 Out #2
4 d1 d1 d1 d1 d1 d1
d1 d1 d1 d1 d1 d1 d1 d1 Out #3
3
6 d1 d1 d1 d1 d1 d1
d1 d1 d1 d1 free token
8 d3
d3 d3 d3 busy token
10 d1 d3 d3 d3 d3
data bit from
d1 d1 d3 d3 d3 d3 d1
#1
12 d1 d1 d1 D3 D3
data bit from
d1 d1 d1 d1 d3 d3
#3
14 d1 d1 d1 d1
d1 d1 d1 d1
16 d1 d1 d1 Full slot
d1 d1 d1 d1
18 d1 d1 d1 d1
d1 d1 d1 d1
20 d1 d1 d1 d3
12/21/2003
d1 d1 d1 d1 Dr. Ashraf S. Hasan Mahmoud
d3 d3 d3 117
Performance of Slotted Ring
• Assumptions
• All stations are identical load-wise
• Arrival process ~ Poisson with λ packets / second /
station
• There are M stations
• Channel bit rate, R bits / second
• Station latency = B bits
• τ is the total ring propagation time
• τ’ is the total ring latency = τ + MB/R
• Packet size: exponential – average packet size =
Xavg bits / packet
• The minipacket length is much less than the packet
size
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 118
60. Distributed M/M/1 Queue
• Since the time for a slot to circulate the
ring (opportunity to transmit) is very
small compared to the packet
transmission time modeled as a
distributed M/M/1 queue
• Arrival rate = Mλ
• Service time = Xavg/R
• From station perspective: effective channel
rate = R/2 – caused by the strategy to
prevent ring hogging
• R is used to compute overall throughput
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 119
Distributed M/M/1 Queue – cont’d
• Network throughput, S is given by
S = Mλ Xavg/R Remember:
Average transfer delay for M/M/1 (total delay) is given by
• For slotted ring: E[service time] Xavg/R
• ρ S T = --------------------- = -------------
1–ρ 1–ρ
• R R/2
• T T + τ’/2 where ρ is the utilization of the queueing system
• Therefore, for slotted ring, transfer delay, T is given by
2 Xavg τ’
T = --------- ----- + ----
1–S R 2
• Result valid for arbitrary packet length distribution
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 120
61. Refined Results
• The previous model does not account for the
huge overhead in each minipacket!!
• Let the minipacket or slot size be Lh (overhead
bits) + Ld (data bits)
• Define h = Lh / Ld
• Using the above definitions, one can write
τ’ = m(Lh + Ld)/R + g = τ + MB/R
where m is the number of slots on the ring and g is the
gap in seconds
• Therefore:
X (1 + h) X
S (1 + h) S
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 121
Refined Results – cont’d
• Substituting in the previous result, yields
2(1+h) Xavg τ’
T = ------------- ------ + ----
1 – S(1+h) R 2
• Now – maximum throughput = 1/(1+h)
or Ld/(Lh + Ld) – which is the correct
result
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 122
62. Example 8:
• Problem: A slotted ring is 1 kilometer long, has 50
stations attached and has a bit rate of 10 Mb/s. Each slot
contains 3 bytes of data, a source byte, a destination byte,
and another byte that includes the monitor and indicator
bits. It may be assumed that each station latency is 1 bit
A) How many slots this ring hold without adding any artificial
delays? What is the gap time? If packets of length 1200
bits are to be transmitted on this ring, find the mean
transfer delay when packets arrive at each station at a
rate of (i) 1 packet / second (ii) 40 packets / second
B) Increase the number of station on the network to 100. (i)
How many slots can the ring now hold without adding
artificial delays? (ii) What is the gap time? Again, evaluate
the mean transfer delay for the same arrival rates and
same packet length.
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 123
Example 8: solution
A) For M = 50 stations
Propagation delay, τ = 5 µsec
Ring latency, τ’ = τ + MB/R
= 5 + 50X1/10 = 10 µsec
Slot length, = 6 bytes or 48 bits
Since τ’ = m(48)/10 + g = 10
Therefore, m ≤ 2 – if m = 2, then g = 0.4 µsec
h = Lh/Ld = 24/24 = 1
Xavg / R = 1200 / 10 = 120 µsec
(i) S = MλXavg/R = 50X1X120X10-6 = 0.006
2(1+h) Xavg τ’
T = ------------- ------ + ---
1 – S(1+h) R 2
2X2
= ---------------- 120 + 10/2 = 490.8 µsec
1 – 0.006 X 2
(ii) S = MλXavg/R = 50X40X120X10-6 = 0.24
T = 928.1 µsec
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 124
63. Example 8: solution – cont’d
B) For M = 100 stations
Propagation delay, τ = 5 µsec
Ring latency, τ’ = τ + MB/R
= 5 + 100X1/10 = 15 µsec
Slot length, = 6 bytes or 48 bits
Since τ’ = m(48)/10 + g = 10
Therefore, m ≤ 3 – if m = 3, then g = 0.6 µsec
h = Lh/Ld = 24/24 = 1
Xavg / R = 1200 / 10 = 120 µsec
(i) S = MλXavg/R = 100X1X120X10-6 = 0.012
2(1+h) Xavg τ’
T = ------------- ------ + ---
1 – S(1+h) R 2
2X2
= ---------------- 120 + 15/2 = 499.3 µsec
1 – 0.012 X 2
(ii) S = MλXavg/R = 100X40X120X10-6 = 0.48
T = 12007.5 µsec
12/21/2003 Dr. Ashraf S. Hasan Mahmoud 125