Research On The Theory And Algorithms Of Cross-Layer Design And Optimization For Wireless Multihop Network

Wireless multi-hop networks have advantages of self-organization, low cost and easy deployment, and can be used to extend coverage and increase capacity of the networks. There are a lot of challenges in the design of wireless multi-hop network due to broadcast property and dynamic wireless link and topology. Traditional layering method is not suitable for the design of wireless multi-hop networks, while cross-layer design method allows information exchanges between different layers to adapt to dynamic wireless environment, and thus achieve optimization of the overall performance of the networks.This dissertation investigates the several unsolved problems in cross-layer design and optimization of wireless multi-hop networks with different scenarios and requirements in their applications by adopting the framework of optimization theory and related methods, and proposes corresponding solutions. These problems include the complexity of algorithms, tradeoff between performance metrics in different layers, time-variability of wireless channel, distributed implementation and integration of new technologies and so on.Cooperative diversity can obtain the gain of the space diversity by allowing nodes in a wireless network sharing their resource and relaying information for each other. This dissertation investigates the power allocation and subcarriers assignment for the downlink of a cellular network with cooperative relay stations (RSs) and OFDMA technology. The problem can be formulated as a mixed integer and nonlinear programming problem. To reduce the complexity while take the advantages of cooperative diversity and multi-user diversity, the problem is divided into two sub-problems which are joint subcarrier assignment and RS selection sub-problem and power allocation sub-problem. By the optimality conditions of each sub-problem, corresponding joint subcarrier assignment and RS selection algorithm and iterative power allocation algorithm for RSs and BS are obtained. Numerical simulation shows that both algorithms can not only have low complexity, but also fully utilize the benefits of cooperative diversity and multi-user diversity to obtain higher performance than the algorithms of assigning subcarriers randomly and allocating power averagely.Uplink transmission in cellular networks consumes most part of transmission energy of mobile stations (MSs). The deployment of RSs can shorten the transmission distance for MSs, and thus reduce their power consumption. There are also close relations between power efficiency and Quality of Service (QoS) of data flows from upper layers. This dissertation develops a framework and algorithms for the cross-layer optimization and resource allocation for power efficiency of MSs in uplink transmission of relay-based OFDMA cellular networks. For inelastic data flows, the framework minimizes uplink transmission power while guarantees their minimum QoS requirements. For elastic data flows, the framework strikes optimal tradeoff between transmission rate and power consumption. Due to time-sharing property of multi-carrier systems, the two problems can be solved optimally by using dual decomposition and subgradient method. The number of subproblems by dual decomposition is proportional to the number of subcarriers, and thus the overall complexity is much lower than solving the primal problems directly. Subgradient method is guaranteed to converge to the optimal solution of corresponding problems. The details of the developed cross-layer optimization algorithms are given. Simulation results show that significant benefits of deploying multiple fixed relays in OFDMA cellular network in reducing power consumption, increasing service rate and saving energy for the uplink transmission of MSs can be obtained through the proposed cross-layer optimization framework and algorithms.Wireless-infrastructured cellular networks have the advantages of self-organization and low cost of deployment. This dissertation proposed a model of multi-hop wireless-infrastructured cellular network with time-varying OFDMA channels and allowing multi-receiver scheme, and distributed cross-layer control algorithm of joint congestion control, routing and resource allocation in such kind of network. The cross-layer design problem for end-to-end transmission is formulated as nonlinear convex programming problem. To obtain a distributed algorithm, dual decomposition method is used to solve the primal problem. Due to the time-varying OFDMA channels, the dual problem can be transformed into a stochastic convex optimization problem which can further be solved by quasi-gradient method. Distributed joint cross-layer optimization algorithm for end-to-end data transmission is obtained. The algorithm makes decision of congestion control, routing and resource allocation according to the channel state in the current time slot and does not need the statistical property of the time-varying channel. In addition, optimal subcarrier assignment requires centralized algorithm with high complexity. To obtain a fully distributed algorithm, a distributed subcarrier assignment algorithm is proposed. The simulation results show that the proposed distributed cross-layer control algorithm can guarantee the convergence of the data rates of the end-to-end flows and stability of the queues in the network. In addition, the algorithm can fully utilize the diversity capacity of multi-receiver scheme and obtain higher performance than using single-receiver scheme.This dissertation investigates optimal cross-layer design of wireless multi-hop network with network coding. Network coding can increase the utilization of wireless resource and thus performance of wireless multi-hop networks by utilizing broadcast advantage of wireless medium. The complexity of coding operation is high when using queuing model of one queue per commodity. An alternative capacity region is formulated by introducing virtual flow variables. The optimal cross-layer design problem subject to this capacity region is solved by dual decomposition method. From the relation of network queues and dual variables, a new queuing model that can reduce the operational complexity of optimal coding is derived. Then, a back-presure-based cross-layer optimization algorithm of joint congestion control, routing, scheduling and network coding is proposed based on the dual solution and the developed queuing model. The stability and asymptotical optimality of the designed cross-layer algorithm are analyzed theoretically. Simulation results further verify the stability and convergence of the algorithm, and also show that with the proposed joint optimization algorithms, network coding can interact adaptively and optimally with other components (i.e., congestion control, routing and scheduling) in network layers and thus achieve optimal performance.