Path diversified multi-QoS optimization in multi-channel wireless mesh networks

Wireless mesh networks (WMNs) have become a promising solution for quick and low-cost spreading of Internet accesses and other network services. Given the mesh topology, multiple paths are often available between node pairs, which thus naturally endorse path-diversified transmission. Unfortunately, like in wired networks, discovering completely disjoint paths in a WMN remains an intractable problem. It indeed becomes more challenging given the interferences across wireless channels in a WMN, not to mention that applications may demand heterogeneous QoS optimizations across different paths. The availability of multiple channels in advanced WMNs however sheds new lights into this problem. In this paper, we show that, as long as the best channels with different QoS metrics are not overlapped between neighboring node pairs, complete disjoint paths with heterogeneous QoS targets are available in a multi-channel WMN. We present efficient solutions to discover such paths, particularly for bandwidth- and delay-optimization. We also develop novel algorithms for accurately estimating path bandwidth and delay in the multi-channel environment. These lead to the design of a practical protocol that extends the classical Ad hoc On-demand Multi-path Distance Vector (AOMDV). Through extensive simulations, we show that our protocol yields significant improvement over state-of-the-art multi-path protocols in terms of both end-to-end throughput and delay.