Cross-Layer Design of Wireless Multihop Random Access Networks

Joint design of flow control, multipath routing, and random access control is considered for wireless multihop networks. Based on a network utility maximization formulation, Aloha persistence probabilities are optimized together with multicommodity end-to-end rates and per-link flow rates. Although the joint optimization of Aloha and flow control was previously tackled using a convex reformulation, adding the routing component renders the problem inherently nonconvex. To cope with this challenge, a successive convex approximation approach is taken to obtain a locally optimal solution efficiently. A parallelized distributed algorithm is developed, which scales well in the network size and exhibits low computational complexity. An online implementation is also proposed and tested. Numerical examples verify the novel design and highlight the performance advantage over state-of-the-art alternatives.