Wireless Reducing Overall Delay In Multi Radio Woban


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Wireless Reducing Overall Delay In Multi Radio Woban

  1. 1. For the Proceedings of the Asia Pacific Advanced Network Reducing overall delay in MULTI-RADIO WOBAN with least per node processing overhead on data packet Asad Ali, Irfanullah, Savera Tanveer, Dr.S.M.H Zaidi,Fouzia Noureen School of Electrical Engineering and Computer Science, National University of Science Technology, Islamabad, Pakistan. Email:{09msitaali,09msitirfanu,savera.tanveer,Drzaidi,08msitfnoureen}@seecs.edu.pk bandwidth hungry applications, the needAbstract-- Wireless-Optical Broadband for such a network that support enormousAccess Network (WOBAN) is a hybrid networktechnology. The back-End of the WOBAN being data and that has high capacity andoptical has very high performance both in bandwidth is apparent. Also opticalterms of speed and bandwidth. There has been network provides high data rate, lowlot of research and numbers of protocols were transmission loss, and low bit error rate,designed and numerous algorithms have been Ultra-high bandwidth to meet the growingproposed to bring the performance of the demand, improved performance, andFront-End at par with that of the optical part. increased transmission rate. On the otherSo in this paper too, we propose a technique hand, wireless network has also greatlyto upgrade the performance of the wireless influenced our lives. They provide numberpart so that there may be lesser processing on of advantages like mobility, flexibility, lesserthe actual data packet and may move cost. So Wireless Optical Broadband Accesssmoothly across the nodes in the wireless partof WOBAN. Also when the data packet reaches network (WOBAN) combines the bestthe Optical Network Unit (ONU), it may be feature of the wireless and opticalforwarded as soon as it reaches the ONU networks. WOBAN is the hybrid of thewithout having to wait for the designated time Wireless and optical networks. WOBAN wasslot. In this way, there will be no time slot proposed so that there may be a flexiblesynchronization delay at ONU . solution to the increased demand of access network. Not only to overcome the demand but also to have cost-effective deployment of a WMN while having higher performanceKeywords: Protocol, Network Technology, due to the optical network. Though theWireless-Optical Broadband Network, delay, communication scenarios [1] of both areONU totally different from each other. Optical network are used for high bandwidth and 1. Introduction for long-haul communication and wireless to provide flexible connectivity. So in There has always been an increased in the WOBAN we combine both in an efficientdemand of the bandwidth. Also there is a and cost-effective manner in order to taketremendous increase in the access network the advantages of the best features of both.traffic and particularly with the introduction Rest of the paper is organized as follows.of services like high definition TV, Video Section 2 highlights the Architecture ofConferencing, telemedicine and other 1
  2. 2. For the Proceedings of the Asia Pacific Advanced NetworkWOBAN, section 3 shows the a typical a Central office (CO) the Optical Linescenario for data communication in Terminal (OLT) resides in CO. The CO andWOBAN, section 4 is about the related work the wireless part is connected through aon the problem, section 5 will be on our trunk fiber. At the wireless part, first of allproposed scheme, section 6 consists of a there are gateway routers called ONUs. Thescenario for our proposed scheme and we’ll rest of the wireless nodes are connected todraw conclusion in section 7. Section 8 will these ONUs. The distance between thebe about the future work and than ONU and the CO is almost 20Kms. However,acknowledgement. this distance can be decreased or increased depending upon the requirement and the 2. Architecture of WOBAN number of ONUs to be supported. End- users whether stationary or mobile, areWOBAN is an optimal combination of an connected to the network through theoptical back-end and a wireless front-end wireless nodes. There is an optical splitterfor an efficient access network [1][3]. between the OLT and the ONUs. The opticalArchitecture of a WOBAN consists of splitter, being passive device, so overall thewireless part- The Front-End and the optical architecture is more robust.Part-The Back-End. At the back end, there is Figure 1: Architecture of a WOBAN 2
  3. 3. For the Proceedings of the Asia Pacific Advanced Network 3. Data Communication in with the speed of the optical part, up to some extent. WOBANFor communication in downward direction, 4. Related workthat is, from CO towards ONU, the packet is Number of routing algorithms has beenbroadcast. The optical splitter split the proposed. Traditional routing approachessignal equally among the entire ONUs. The like SPRA and MHRA are both based on theentire ONUs rejects the packet except the shortest path. They don’t consider otherone for which the packet is aimed. In characteristics of the network likeupward direction, when end users want to congestion, available bandwidth, load onsend data packets, it does so by passing the the link etc. Most often the shortest path ispacket to the nearby nodes. That nearby not the best path. So both SPRA and MHRAnode passes the data packet to its other both do not overcome the delay problem inneighboring node nearer to the ONU. In particular and also the load balancing andThis way, packet pass hop by hop to congestion problem. Another approach isdifferent nodes to reach the ONU and then PTRA, chooses the path with the highestfrom ONU, through splitter, ultimately to throughput based on the Link Statethe OLT. From OLT, the data packet is sent Advertisement (LSA). Again, the path within the similar fashion discussed above, that the highest throughput might not be theis, the packet after reaching the OLT is sent shortest path. As we are interested inback over the same fiber trunk. The optical choosing the shortest path having highestsplitter splits the signal equally among the throughput and least delay.ONUs. The entire ONUs will reject thepacket except the ONU in which there is the There are some other routing algorithmstargeted user of the packet. If this packet is proposed to overcome the delay problem.meant for a node out side this WOBAN, it is In DARA [3] and CaDAR [4] the authors haverouted by the CO outside the network. So in calculated the delay of each link and thenwireless part when the data packet is for forwarding the packet, overall delay ofpassed hop by hop, doing various kinds of the end node from the ONU is calculated.processing and making routing decision at The authors have identified four types ofeach hop. There comes a significant delay. delay.Since optical part has high speed, high 1. Transmission delaybandwidth and high capacity, thus havinghigh performance. So the performance ofthe wireless part must be brought at par 2. Slot synchronization delaywith that of the optical part. So first all thisdelay must be reduced, in order to match 3
  4. 4. For the Proceedings of the Asia Pacific Advanced Network 3. Queuing delay Routing Dela y Good Load algori thm handled? throughput? Balancing? SPRA Upto some Not so much no extent 4. Propagation delay—Nodes being MHRA Not No so much no completel y closer to each other. So negligible. PTRA Not Yes Yes Table 1: Compa rison of va rious algori thms DARA Yes* Yes Yes Ca DAR Yes* Yes YesTable 1 shows the comparison of various Table 1: Comparison of Various Algorithmalgorithms. It is obtained by studyingvarious algorithms and their characteristics 5. Proposed schemeabout whether they solve the delay andthroughput problem or not. The above The link condition and the capacity requiredtable shows that the DARA and CaDAR are for a node is advertised in LSAs. LSAs arealmost same. However, CaDAR is better advertised periodically. So if after a LSA, linkthen DARA in a sense that it also assigns condition are changed, say for instance,capacities to all the links. In both the flow on the link is changed, then the linkproposed algorithm proposed, though they will be either under-utilized or over-utilizedvery efficiently calculate the link delay, and and may lead to congestion, packet lossselects the link with the least delay. depending upon the whether the demandHowever, calculating these so many types of the node was increased or decreasedof delay for a link and for each packet and after that LSA.. Time-Slots for various nodesthen for the whole path, may itself be a and their data is adjusted also according todelay. If a data packet has to undergo these the LSA. So all the algorithm discussedso many processing and decision making, above, heavily depends upon the LSA, thusthe reduced delay may not be significant. So in the inter-LSA period, the networkin this paper we propose a mechanism in condition may change dramatically, thuswhich there will be reduced processing and leading to the wrong estimation anddecision making on the actual data packet. calculations. Also the slot may be wronglyThough the shortest path with the least synchronized. And due to lack of networkdelay is calculated in similar fashion as in information the capacity may be wronglyCaDAR and DARA but however, the assigned to the link thus leading to otherprocessing overhead will be much reduced offensive results like congestion, packet losson the actual data packet, so it will move and so on. One solution to all these is tosmoothly across the nodes. Thus there will send LSA quite often but in this case abe significant reduction in the delay. major chunk of bandwidth will be used by sending LSAs themselves. So In our proposed scheme, we will send a control packet ahead of the actual data packet. Thus there will be no need of LSAs. In this 4
  5. 5. For the Proceedings of the Asia Pacific Advanced Networkmethod, we will embed the sequence will be forwarded as it reaches the node.number and size of the data packet in the Similarly time slot will be reserved at thecontrol packet. The control packet is ONU, so as data reaches the ONU, it will beactually the header of the data packet forwarded without waiting for theseparated from the data packet. So for designated time slot, as the designated timesynchronization and identification purpose slot will reach as soon as the data packetwe use the sequence number. The control will reach the ONU.packet will contain fields for the sequencenumber and the size of the data. One extra 6. Scenariofield will have to be added in the datapacket for the sequence number using thebits of reserved field. The data packet willtravel ahead of the data packet, thus findingthe shortest path with the least delay forthe actual data packet. The data packet willfollow the same path which was followedby the control packet. All the routingdecision will be made and the shortest path Figure 2: Back-End of WOBANwith the least delay will be calculated for Consider the Fig 2, above. Consider thatthe control packet only. There will be no End-User 2 generated data, suppose theprocessing on the actual data. The sequence number of the data be 500. Nowintermediate node, just looking at the this sequence number and the size of thesequence number of the packet will data packet will be embedded in the controlforward it on the same path on which the packet. The only thing which will becontrol packet was forwarded. Also the included in the data packet will be a fieldcontrol packet will tell each node exactly for sequence number, so that thehow much data is coming from a particular intermediate node recognizes this sequencenode. So in this way, capacities will also be number. Now suppose the control packetassigned to the link according to the after reaching node A (since End-User isinformation of the control packet as it connected to the network via node A), findscontains the size of the upcoming data out the shortest path with the least delaypacket. Thus not only there will be no link to node B, at B it find the shortest withprocessing on the data packet and least delay link to node C and then to ONU.capacities will be assigned to the links, but Based on the control packet, capacities willalso the control packet will so synchronize also be assigned to the links {AB} and { BC}.the time slot at each node, so that the Also the intermediate node B and C willwhen the data reach the node, it will not adjust the time slot according to the arrivalhave to wait for designated time slot and of the data packet and then will not make 5
  6. 6. For the Proceedings of the Asia Pacific Advanced Networkany processing on the actual data packet, soon as its data packets reaches the node’sthe only things which the intermediate queue, its time slot will also come, so it willnode will do is to check the sequence be forwarded without having to wait for itnumber. Thus recognizing the sequence designated time slot. Again, the only thingnumber, they will forward the data packet which the intermediate node has to check ison the path on which the control packet the sequence number of the data packet.was followed. Another advantage of the The packet will be forwarded in similarcontrol packet is that, when the control fashion at ONU as well.packet reaches the node B and C, it will alsoso synchronize the time slots at the B and C,that as soon as the data packet arrive thesenodes, its time slot also come. So the datapacket will be forwarded as soon as itreaches the node without having to wait fordesignated time slot and thus there will beno queuing delay and time slotsynchronization delay as well. Similarly at Figure 3: Queue of packets and their Time-SlotsONU too, the control packet will reserve thetime slot for the data packet and will alsoreserve the bandwidth for the data packetto be forwarded on the fiber trunk. So 5. Conclusionagain as soon as the data packet reaches We have shown a fine technique to reducethe ONU, its designated time slot will also the overall delay and efficiently use thecome, so it will be forwarded without network resources. With DARA and CaDAR,having to wait for designated time slot. In delay will have to be calculated for all thethis way three types of delays namely, links and ultimately for the entire path, forprocessing over head delay on the data each packet sent and then decision willpacket, queuing delay and time slot have to be made regarding the shortestsynchronization delay, significantly path with the least delay for every dataimpacting the overall performance of the packet. Also both the algorithm dependswireless part are reduced. upon the LSA, since network condition mayFigure 3 below shows data packet of various change even before next link stateEnd-Users and their time slots. Packet will advertisement. So in this paper, we havebe forwarded in the time slots one by one. shown an efficient mechanism in assigningSince we know that time slots are capacities to the link and synchronizing thesynchronized on the basis of the LSAs. The time slots of the nodes and a mechanism incontrol packet of a particular data packet which there will be no processing or routingwill so synchronize the time slot that as decision on the data packet. The only thing 6
  7. 7. For the Proceedings of the Asia Pacific Advanced Networkwhich the intermediate node has to check is the queue for its designated time slot. Thusthe sequence number of the data packet, there will be significant reduction in theand it will be automatically forwarded on overall delay. Overall all the three types ofthe path on which the control packet was delays will be eliminated, namely, queuingforwarded. Since there is no other delay, time slot synchronization delay andprocessing on the data packet and also the processing delay. The need for LSAs are alsocontrol packet will also reserve the time for eliminated. So since there are no LSAs, thusthe data packet at intermediate nodes and there will be an efficient utilization ofONU, so the packet does not have to wait in resources. Table 2: Comparision of various algorithms with the proposed scheme 6. Future Work So that there may be equal load on the link and a sort of load balance may prevail inSome time it may happen that a single link the network. Also in the link, capacities areis over loaded, the traffic of all the node sometime not properly assigned. So we alsoflow on the same link or path, which may intend to work on equally assigning thelead to failure of the link, packet loss, low capacities to links. Though there wouldthroughput or any sort of consequences. So have been worked carried out in this area,in future we intend to work on equally but yet we will try to refine the existingdistributing the traffic load on all the links. 7
  8. 8. For the Proceedings of the Asia Pacific Advanced Networkwork or to propose new methods of equally [6] N. Correia, J. Coimbra, and G. Schützassigning the capacities and load balancing. “Fault-Tolerance Planning in Multiradio Hybrid Wireless-OpticalReference: Broadband Access Network” VOL. 1, NO. 7/DECEMBER 2009/ J. OPT. [1] N. S. C. Correia†, J. Coimbra†, G. COMMUN. NETW. [7] E. S. Son, K. H. Han, J. H. Lee, and Y. C. Schutz “Multi-Radio Hybrid Wireless- Chung, “Survivable network Optical Broadband Access Networks” architectures for WDM PON,” in PTDC/EEA-TEL/71678/2006. Optical Fiber Communication [2] Abu (Sayeem) Reaz, Vishwanath Conference, 2005. Ramamurthi, Suman Sarkar, Dipak [8] Vivek P. Mhatre, Henrik Lundgren, Ghosal, and Biswanath Mukherjee Christophe Diot “MAC-Aware Routing “Hybrid Wireless-Optical Broadband in Wireless Mesh Networks” AccessNetwork (WOBAN): Capacity [9] Seung-heub Jeo1, Young-bok Cho1, Enhancement for Wireless Sang-ho Lee “Mobility Mechanism Access”978-1-4244-2324-8/08/ © included Mobile Routing Method in a 2008 IEEE "GLOBECOM" 2008 Wireless Mesh Network Environment” proceedings. 9781-4244-3428-2/09/ ©2009 IEEE [3] Suman Sarkar, Hong-Hsu Yen, Sudhir [10] Douglas S. J. De Couto Daniel Dixit, and Biswanath Mukherjee Aguayo John Bicket Robert Morris “A “DARA: Delay-Aware Routing HighThroughput Path Metric for Algorithm in a Hybrid Wireless-Optical MultiHop Wireless Routing” ACM Broadband Access Network (WOBAN)” 1581137532/03/2009 IEEE Communications Society ICC 2007 [11] Bo Rong, Member, IEEE, Yi Qian, proceedings. Senior Member, IEEE, Kejie Lu, Senior [4] Abu (Sayeem) Reaz1, Vishwanath Member, IEEE “Enhanced QoS Ramamurthi1, Suman Sarkar, Dipak Multicast Routing in Wireless Mesh Ghosal1, Sudhir Dixit2, and Biswanath Networks” IEEE TRANSACTIONS ON Mukherjee “CaDAR: an Efficient WIRELESS COMMUNICATIONS, VOL. 7, Routing Algorithm forWireless-Optical NO. 6, JUNE 2008 Broadband Access Network” 978-1- 4244-2075-9/08/ ©2008 IEEE Communications Society ICC 2008 proceedings [5] Suman Sarkar, Student Member, IEEE, Sudhir Dixit, Senior Member, IEEE, and Biswanath Mukherjee, Fellow, IEEE “Hybrid Wireless-Optical Broadband – Access Network (WONBAN) :A Review of Relevant Challenges” JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 25, NO. 11, NOVEMBER 2007. 8