Cross-layer design has been a hot research area since 1990s. By exploiting interactions of the protocols in different layers, network performance can be improved significantly. The mentality of ”Layering as Optimization Decomposition” has provided a unified framework of cross-layer design in wireless communication networks. In this thesis, a cross-layer design framework, which aims to minimize total power consumption in a power and bandwidth limited multihop wireless network, has been exploited....[ Read more ]

Cross-layer design has been a hot research area since 1990s. By exploiting interactions of the protocols in different layers, network performance can be improved significantly. The mentality of ”Layering as Optimization Decomposition” has provided a unified framework of cross-layer design in wireless communication networks. In this thesis, a cross-layer design framework, which aims to minimize total power consumption in a power and bandwidth limited multihop wireless network, has been exploited.

In the cross-layer design framework, we formulate a joint routing and physical layer resource allocation problem. In the existing literature, dual decomposition technique is usually employed to solve cross-layer optimization problem in a distributed way. However, in our scenario, we add node power constraint and network total bandwidth constraint to the optimization problem, which changes the problem structure. Standard dual decomposition technique does not work. Therefore we introduced multilevel decomposition. In the upper level, dual decomposition is employed to decompose the cross-layer problem into network layer subproblem and physical layer subproblem. Network layer subproblem can be solved by dual decomposition directly. In physical layer subproblem, the objective function is nonlinear and non-strictly convex. Primal decomposition technique is used to decompose physical layer subproblem into a per-node based physical layer subproblem. After introducing auxiliary variables, we prove that the optimal solution of per-node physical layer subproblem lies on the boundary of a convex hull of a specific set, which is the union of epi-graphs of a family of functions. Finally the joint routing and physical layer resource allocation problem is divided-and-conquered in each subproblem and can be solved distributedly. Simulation results show that our proposed algorithm reduces total power consumption of the network significantly.

After routing and resource allocation are jointly optimized by our proposed technique, cooperative communication is introduced in the physical layer to further reduce power consumption in the multihop wireless network. To keep the distribute property of the optimization algorithm, direct transmission and cooperative communication is dynamically chosen on each link locally. We propose an algorithm to select the suitable relay node and the optimal power allocation among source and relay nodes. Simulation results show that cooperative communications can further reduce the total power consumption in the network protocol.