Abstract — The performance of conventional gateway access optimization techniques deteriorate dramatically when traffic load is dynamic. In this article, we propose a novel gateway access algorithm called ‘Cog Gap’, which is a cognitive method and is designed for Wireless Mesh Networks to maximize the network utilization. The proposed Cog Gap utilizes a destination-hub access model, where multiple gateway nodes are connected by wired links, and packets from a source node can be sent from any connected gateway nodes to increase transmission opportunities. In Cog Gap, we use the Hidden Markov Model (HMM) and the expectation maximization method to handle uncertain traffic pattern and the loss of probing results. A traffic allocation algorithm is then proposed to optimize dynamic multi-gateway access. By modeling the route state determination and transition, we transform the opportunistic gateway access problem into a Markov decision process (MDP) problem. A heuristic and adaptive algorithm named hindsight optimal is used in solving MDP. Simulation results have proven that the proposed Cog Gap algorithm can make full use of the transmission opportunities and does not incur noticeable protocol overhead.

Abstract—Wireless mesh networks are popular as a costeffective means to provide broadband connectivity to large user populations. A mesh network placement provides coverage, such that each target client location has a link to a deployed mesh node, and connectivity, such that each mesh node wirelessly connects directly to a gateway or via intermediate mesh nodes. Prior work on placement assumes wireless propagation to be uniform in all directions, i.e., an unrealistic assumption of circular communication regions. In this paper, we present approximation algorithms to solve the NP-hard mesh node placement problem for non-uniform propagation settings. The first key challenge is incorporating non-uniform propagation, which we address by formulating the problem input as a connectivity graph consisting of discrete target coverage locations and potential mesh node locations. This graph incorporates non-uniform propagation by specifying the estimated signal quality per link. Secondly, our algorithms are the first to minimize the number of deployed mesh nodes with constant-factor approximation ratio in the nonuniform propagation setting. To achieve this, we formulate the Degree-Constrained Terminal Steiner tree problem and present approximation algorithms which leverage prior results on the Steiner tree problem. Third, it is impractical to measure all possible potential mesh links, and therefore deployment planning must rely on estimations. To address this challenge, we extend our algorithm to iteratively measure the links in the solution Steiner tree, refining the graph input on a per-link basis in order to ensure the deployed network is not disconnected. Finally, we use propagation measurements at 35,000 locations in the deployed GoogleWiFi network to investigate placement in a realistic, non-uniform propagation environment. Under this measured propagation setting, our algorithms result in up to 80% fewer mesh nodes than current algorithms and only require an average of 3 measurements per deployed mesh node to ensure backhaul connectivity. I.

This work proposes the Wireless-mesh-network Proactive Routing (WPR) protocol for wireless mesh networks, which are typically employed to provide backhaul access. WPR computes routes based on link states and, unlike current routing protocols, it uses two algorithms to improve communications in wireless mesh networks taking advantage of traffic concentration on links close to the network gateways. WPR introduces a controlled-flooding algorithm to reduce routing control overhead by considering the network topology similar to a tree. The main goal is to improve overall efficiency by saving network resources and avoiding network bottlenecks. In addition, WPR avoids redundant messages by selecting a subset of one-hop neighbors, the AMPR Corresponding author.