Study Of The Performance For Delay-Sensitive Services In Wlan Over OBS Systems

Next generation wireless networks are characterized heterogeneously in term of the underlying technology to satisfy users'demands of high bit rate services at low cost and low delay with seamless mobility at anywhere and at anytime. Recent advances both in wireless and optical communications fields provide some maturating technologies to achieve such goal. On one side, wireless local area network (WLAN) has emerged as a new technology with ability providing relatively high bandwidth at comparatively low cost of construction and suitable level of freedom for roaming users. On the other side, optical burst switching (OBS) has also emerged as a promising candidate to transport tremendous data packets with high processing speed and network resource efficiency. Therefore, designers may take the advantages of both fields to achieve the goal of the next generation wireless communications. The combination of the two technologies is expected to provide better services for users. OBS with its advantages for packet services in particular may be a suitable candidate to serve as the backbone network to extend WLANs services. Therefore, WLAN over OBS is one of the possible solutions for the next generation wireless networks. However, there are some issues for such systems that are to be solved. In addition to the ones related to WLAN itself such as the throughput and delay performance, features and problems raised from the employment of the OBS backbone should be studied in order for such system to serve for next-generation with extended coverage and transportation ability.This dissertation is conducted to analyze the key characteristics, such as the throughput, packet delay and packet loss that are required for designing the WLAN over OBS system to serve for delay-sensitive applications. The study is carried out from three aspects: the individual WLANs, the individual OBS networks and then the achieved results are used for combining the two parts of the system, i.e., the WLAN over OBS system.The dissertation starts with analyzing the characteristics of the backbone network according to the adopted protocol. In addition to investigate the OBS that utilizes one way reservation, like JIT (just in time) and JET (just enough time), OBS backbone using the two way reservation scheme (wavelength routed OBS, WR-OBS) is also studied. The latter protocol has the advantage of low burst loss rate in comparison with the former ones, so it has been paid more attention within the study. Of course, two-way reservation requires more waiting time than the one-way reservation, so we investigate them in terms of the delay-sensitive application by analyzing the throughput and packet delay. Burst aggregation in WR-OBS can be done through two methods: limited size burst (LSB) and non limited size burst (NSB). Depending on the used aggregation method, maximum network traffic load is influenced by lightpath load. Simulations have been carried out for two network diameters: 500Km and 1000Km. The reuse factor is investigated to characterize OBS networks with dynamic wavelength assignment. The results show that the system can offer better performance with increasing core-to-edge bit rate ratio. Delay (waiting) at edge router, in general, significantly influences the performance of the system. Increasing packet interval and using fast core router can attain low probability of burst blocking.Then, the throughput and delay of WLANs, crucial parameters due to the limited sharing transmission bandwidth, are studied for the two DCF (distribution coordination function) access mechanisms in IEEE 802.11 protocol: two-way and four-way handshaking. In this study, two cases of the wireless transmission channels are considered: ideal and error-prone channels. An analytical model is developed to examine characteristics of the terminal accessing networks (WLANs): throughput, packet delay, probability of packet dropping and time to drop a packet. The model takes into account influences of: (1) the number of retransmissions, (2) the error level in the utilizing transmission channel, (3) type of the transmission channel mechanism, and (4) the size of the network (number of stations under one access point). The achieved results have been compared with other models for validation purpose. The results show that the terminal networks that use request to send/clear to send (RTS/CTS) (i.e. four-way handshaking) mechanism have a good immunity against the error over the available range of retry limit; the throughput is also unchangeable with the size of the network over the range of retransmission limit. For the networks using basic access (i.e. two-way handshaking) mechanism, the throughput is suppressed over all the range of the retry limit when the amount of errors in the transmitting channel increases. It is also quite sensitive to the size of the network. However, the throughput does not change with the retry limit when it exceeds the maximum number of the backoff stage in DCF mechanisms. In the both mechanisms, the probability of dropping a packet is a decreasing function of the number of retransmissions. The time to drop a packet in the queue of a station is strongly dependent on the number of retry limit, the size of the network, the utilizing medium access mechanism and the amount of errors in the channel. Comprehensive comparisons between the two mechanisms are conducted for these aspects. The results show that the amount of the channel bit error rate (BER) and the size of the network are essential to predict throughput and the packet delay characteristics for each adopted mechanism. When the BER increases to a certain level, few packets can be successfully received and the number of retransmissions increases, which leads to degradation of the throughput and the packet delay. According to the achieved results in this study, the network designers should make the required compromise between the throughput and the packet delay by selecting the optimal length of the transmitted packets. We present a novel approach to adaptively determine the packet length according to the circumstance of the utilizing channel. The results show that the packet delay can be kept within a suitable level (according to the standard) for relatively high noise level channel and a large number of stations.After analyzing WLANs and OBS networks, we investigate the performance of the WLAN over OBS system. Simulation study for some characteristics of the whole system, such as the throughput and delay, is performed using the Network Simulator-2 (NS-2). The connection between the two different protocols in OBS and WLANs is validated by NS-2 simulation and simple test-bed. End-to-end delay for the WLAN over OBS using one-way protocol and two-way protocol in the backbone is modeled and analyzed. The results indicate that the wireless LANs play a bottleneck for such system and the delay in the backbone is comparatively negligible, so the analysis on the terminal networks consists of a considerable portion in this study. Also, the load influence is presented and we compare it with the other system which uses a wavelength circuit switching network as the backbone for the WLANs. The results indicate that the throughput in the case adopting OBS to serve as a backbone is superior above 20% to that in the case using the wavelength switching system. At last, the handoff issue in the WLAN over OBS is addressed.In short, this study offers the essential characteristics as a guideline for designers who are interested in build wireless accessing systems: the OBS, WLAN and WLAN over OBS. From the achieved results, it can be shown that WLAN over OBS system helps to seize the advantages of optical and wireless technologies to support the delay sensitive applications when the related parameters and mechanisms of the system are chosen carefully.