Research On MAC Protocol In UWB Based High Rate Wireless Personal Area Network

UWB-based high rate wireless personal area network (UWB HR-WPAN), which integrates the appealing UWB technology, is the inevitable outcome to meet the increasing demand of person-centered high rate interconnections. It has drawn much attention from academic world and industrial world due to its technical advantages and bright applications. Medium Access Control (MAC) protocol is one of the key technologies and basic protocols in UWB HR-WPAN. The research of the MAC protocol plays a significant role in the network performance optimization, application demands adaptation and multimedia service QoS improvement.On account of the differences in network form, transmission technology and application demand, the traditional wireless MAC protocols cannot accommodate to the new features of UWB HR-WPAN. Some open issues remain to be solved, for example, the network signaling is unreliable, QoS guarantee for heterogeneous traffic is weak, the space resouce can't be exploited and the video application can't be supported efficiently. Therefore, oriented to applications such as multimedia interconnection and digital home, the thesis takes an intensive study of MAC protocol and scheduling algorithm in UWB HR-WPAN via cross-layer methodology to achieve high performance and QoS guarantee.First of all, the access control architecture, which is the basis of the MAC protocol in UWB HR-WPAN, should be efficient, self-organized and be able to provide reliable signaling. Therefore, on the premise of the investigation into the state-of-art, the thesis does the following works. On one hand, a PSMA/CA protocol for contention period in the superframe is proposed to address the issue that the traditional CSMA/CA cannot be implemented in UWB networks. On the other hand, the author further proposed a relay based beacon loss recovery scheme RBLR to deal with the inefficiency because of the beacon loss. The scheme exploits the special network structure and makes the beacon recovery using the device relay.Then, concerning the scheduling problem for heterogeneous traffics, the author proposes a dynamic programming based optimal scheduling algorithm for multi-class traffics based on utility functions. The algorithm, which exploits the elastic scheduling architecture, combines the fairness and characteristic of traffic and wireless channel. The dynamic programming method is further employed to solve the optimization problem of maximizing network utility.Next, scheduling based on space diversity is able to further improve the efficiency and QoS further in UWB HR-WPAN. To conquer the defects in terms of collision model construction, reliability and QoS guarantee in the related works, a novel maximum sustaining interference based space division scheduling algorithm MSDS is proposed. The MSDS builds a collision model directly from the UWB signal and operates the optimization in both the time dimension and space dimension.Finally, video is a main applications in UWB HR-WPAN and the design of the scheduling algorithm for the video stream is closely dependent to its feature. The author proposes a reinforcement learning based video scheduling algorithm RLVS. Incorporating the feature of video traffic and wireless channel, the algorithm employs the modern reinforcement learning method with feature abstraction and state generalization to solve the optimization problem with the curse of dimensionality and modeling.The contributions of the thesis are laid out as follows:1. A relay based beacon loss recovery scheme RBLR is proposed. The scheme can improve the beacon success rate and network resource efficiency with little overhead expense. Simulations prove that in the condition that the relay device is sufficient, the RBLR can keep the beacon success rate above 80% while the time overhead in one superframe never exceeds 4%.2. A dynamic programming based optimal scheduling algorithm for multi-class traffics is proposed. The algorithm not only improves the network utility in dynamic wireless channel, but also makes a well balance between optimal utility and fairness. Simulations prove compared to the traditional proportional allocation algorithm, our algorithm can improve the network average utility up to 211.6%. And the fairness index maintains above 0.9 when the flow number is no greater than 20.3. A maximum sustaining interference based space division scheduling algorithm MSDS is proposed. The MSDS not only keeps low overhead and reliable transmission but also enhances the network capacity and QoS guarantee. Simulations prove compared to the traditional TDMA FIFO algorithm, the MSDS can improve the network utility up to 219.5%, while the acknowledgement overhead in one superframe never exceeds 120us.4. A reinforcement learning based video scheduling algorithm RLVS is proposed. In spite of the long training time, the RLVS can make a distinct improvement on the QoS for video traffic. Simulations prove compared to the SRPT and EDF algorithm, the RLVS can reduce the weigh job failure rate (WJFR) up to 23.5% and 22.8% when the flow number is no greater than 10.