WIND: A Tool for Capacity-constrained Design of Resilient Multi-tier Wireless Mesh Networks

In this work we present a tool for the automated design of Wireless Mesh Networks (WMN). The general scenario we envision is that of constructing WMNs with WLAN networks as clients (tier 1) and a mesh network (tier 2) to provide interWLAN as well as gateway connectivity. Our main aim in this work is to a) design capacity-constrained WMNs, b) build resilient WMNs with transient demands. IEEE 802.11 based single-hop WLANs are now widely prevalent [1]. However adhoc deployment of such networks have the following issues: (i) They cannot adequately address QoS-constrained capacity requirements [5] and (ii) Provide cost-efficient backbone connectivity to the AP. Removing wired connectivity to APs is an important goal in order to increase the cost savings accrued by avoiding the deployment of a wired backhaul connectivity. But additionally, a suitable technology is necessary to adequately replace the large bandwidth capability of wired networks. Wireless Mesh Networks (WMN) are gaining popularity as a solution to provide a wireless backbone and address the capacity constraints of a single-hop wireless network [2]. In WMNs, mesh nodes acting as routers are placed in the network to provide the backbone connectivity to the gateways. The networks based on such a mesh backbone topology, allows multi-hop wireless access, support for self-forming and rapid reconfiguration of topologies. For a single-hop network, the benefits are the absence of wired connectivity from the Access Points (AP) to the backbone and the use of multiple radios by the APs to communicate with the end-users and the backbone. WMNs using WiMAX are also anticipated to significantly improve the performance of ad hoc networks, wireless personal area networks (WPAN), and wireless metropolitan area networks (WMAN) [2]. Hence it is important to have mechanisms to automate the design of such networks. While the design of WMNs falls in the same class of network design problems as encountered in wired as well as cellular networks, there is a significant difference in the node capabilities and the associated constraints and cost-functions (link creation cost). For example, the wireless nature of the links (including backbone links between mesh nodes) give rise to cost-functions not encountered in other networks, and also the use of multi-hop wireless transmission results in additional scheduling constraints. Fig. 1. A typical mesh network scenario.