Channel Adaptive MAC Protocol Study With Cross-Layer Design

Recently, as a cheap and convenient wireless local access method, wireless LAN has aroused extensive interests. The wireless media access control (MAC) protocols have also attracted the attention of industrial and academic researches. Although the traditional layered protocol works well in wired networks such as Internet, it may not be efficient for wireless networks when heterogeneous traffic is served over a wireless channel with limited and time-varying capacity and high bit error rate (BER). Efficiently utilizing the scarce radio resources requires a cross-layer joint design and optimization approach. Nowadays, the concept of cross-layer design has been widely accepted. As a result, a large number of cross-layer design proposals have appeared in the recent literature.This dissertation mainly discusses the channel adaptive MAC protocol for wireless LAN with cross-layer design. The main works and conclusions are listed as follows.Firstly, we propose a frame-fragment based network simulation model. For current network simulator, there is a basic assumption that the channel condition keeps constant during the transmission of a frame. However, if the time to transmit a frame is rather long, or the wireless channel varies faster, the channel condition will change evidently, and the simulation results would not keep precise any longer. The basic idea of the proposed model is to divide the frame into several small fragments according to the speed of channel variations, and during the transmission of a fragment, the channel condition is assumed to keep constant. By this model, we can get more precise result with only a little additional simulation time. This model provides a simulation platform for the following studies.Secondly, we propose a mechanism named EACK (Enhanced ACK), which can distinguish collision due to network congestion from transmission error due to hostile channel condition. With only a few modifications to the IEEE 802.11 standard, EACK can be used to kinds of cross-layer design.Thirdly, we study the frame length adaptive strategy under a noisy channel condition. By analyzing the impact of frame length, bit error rate (BER) and the number of nodes on IEEE 802.11 saturation throughput, we find that the optimal frame length maximizing the throughput only depends on BER, while is independent of node number. Therefore, a simple frame length adaptive strategy based on BER range is put forward. Furthermore, we evaluate the performance of IEEE 802.11 fragmentation under noisy and jammed network. Finally, with EACK mechanism, we propose a channel adaptive fragmentation (ADF) strategy, which considers both the collision and transmission error. ADF has the ability to adapt the fragment size to channel condition, even without the knowledge of channel state information (CSI).Fourthly, we investigate channel adaptive multirate MAC protocols. First, an improved multirate ARF protocol named AMARF is presented. AMARF switches the rate based on the received ACK information as ARF does. The difference is that AMARF assigns each data rate a separate success threshold which can be adjusted in an adaptive manner according to running conditions. AMARF protocol can be implemented easily by software, without any modification to the current IEEE 802.11 standard. Extensive simulation also shows that compared with ARF and its variations, AMARF has a distinct performance improvement. Second, we propose another rate adaptive MAC protocol named MEACK, which chooses data rate based on Signal to Noise Ratio (SNR). In MEACK, the ACK frame is used to feed back CSI. The key point of MEACK is to modulate and code the MAC header in a reliable manner, such that MAC header can be received correctly by the destination. The simulation result shows that MEACK reacts to channel variations quickly while having fewer overheads.In the last, we do some research on high rate MAC protocols. We propose a constant duration transmission (CDT) scheme based on MEACK. CDT deals with the problems of maximum throughput limit and average throughput limit effectively, which are aroused when the current IEEE 802.11 is applied to high rate networks. In CDT, the correlation of wireless channel is exploited. The main idea of CDT is that multiple MSDUs (MAC Service Data Unit) are aggregated into a super frame to transmit according to data rate, and the transmission duration of the super frame is kept approximately constant. All the MSDUs in a super frame share a single frame header, while having different frame check sequences, therefore, the receiver can decode each MSDU separately. We validate CDT scheme by analysis and simulation. Furthermore, to improve the performance of CDT in fast time-varying channel, we try a channel prediction method with AR model to reduce the channel estimation error. Simulation shows that this proposed method with a two-order or three-order predictor can increase the throughput remarkably.