Dublin City University ODCSSS Project Summer 2008 Student: Madalina Fiterau Supervisors: Dr. Olga Ormond Dr. Gabriel-Miro Muntean
Aim of the project
Study the performance of handover for a group of mobile devices as they move simultaneously from one network to another
Propose an intelligent and adaptive algorithm to support the movement of the group with the best possible overall QoS
Apply appropriate metrics for the assesment of group welfare in heterogeneous networks
Scenario BS AP BS MSs UMTS WiMAX WLAN
Background Research IEEE 802.21 Media Independent Handover - standard being developed to enable the handover of IP sessions from one layer 2 access technology to another, to achieve mobility of end user devices Y. Wang Handover algorithm with buffer prediction A. Dutta Location assisted proactive handover with pre-authentication and pre-configuration D. Cavalcanti Connectivity opportunity selection algorithm exploring the possibilities of multi-hop networks in BS detection and selection G. Lampropoulos Enunciation of 5 handover principles; proposed architecture conforms to all considered handover principles, using the information, commands and events provided by the 801.21 standard S. Yoo Neighbor information-based predictive handover architecture that would place a handover decision engine between the MIH function and the L3 mobility protocol
What could be improved?
Not all the NSAs considered UMTS in the simulations – or UMTS/802 handovers
Few approaches took into account the situation when multiple users switch from one network to another simultaneously
Even fewer employed performance indicators to evaluate the welfare of the whole group
So far, an algorithm focused on a group of mobile devices hasn’t been developed
Characteristics of algorithm
Focus on multiple users – that might have similar profiles
Verticality of handover: UMTS/802.21
Real-time / non-real time applications
Overcoming difficulties Problem Solution Develop a user-centric NSA that must avoid ping-pong effect, having no information on the number of users on the network Introduce a QoS-based delay and a probabilistic handover approach to the NSA Unify selection mechanisms for UMTS and 802 based networks Correspondencies based on previous simulations and mean, expected or predicted values for certain parameters Adapt selection mechanism for real time and general purpose applications Based on previous research, attach weights to network parameters according to the type of application. Simulate appropriate scenarios Extend the NIST Patch for NS-2,29 Features: UMTS add-on, WiFi, WiMAX, 802.21 MIH, mobility: MIP, SCTP Find relevant metrics to asses the impact of the NSA on group welfare Employ statistic functions (sum, mean, standard deviation) to evaluate overall performance
Overview 802.21 MIH Function Radio Access Technologies L2 Triggers and Events Handover Messages Information Service
Algorithm begins by scanning for available networks.
For each candidate network, compute Quality function based on:
According to the difference in quality between the current and first candidate network, delay handover – to allow users that best fit the network to switch first.
If, after a handover, a serious degradation of network quality is detected, switch network according to a random variable with a probability proportional to the quality degradation.
real- time applications non-real-time applications 802.21 MIH Function Radio Access Technologies L2 Triggers and Events Handover Messages Information Service
Overview MIH Trigger r For Each Scanned Network, compute QoS function. QoS gain above threshold or already delayed? Delay Connect! Ping-Pong Detect Y N Y Random 802.21 MIH Function Radio Access Technologies L2 Triggers and Events Handover Messages Information Service
Does it work? Simulation of an “Always select the WLAN” algorithm. All the nodes communicate to the router through the AP, resulting in a drop of throughput and increasing packet loss. Simulation of the randomized network selection. Part of the nodes will opt for the BS, therefore lessening the strain on the WLAN – reducing packet loss and increasing throughput. AP BS BS AP Router Router
Increase in throughput Throughput (kB/s) Time (s) Readings for the first node that stayed within the WLAN
Increase in throughput Throughput (kB/s) Time (s) Readings for the second node that stayed within the WLAN Packet loss at the WLAN “ Always WLAN” Randomized NS 365 Pks 210240 Bytes 80 Pks 46080 Bytes