ActMesh- A Cognitive Resource Management paradigm for dynamic mobile Internet Access with Reliability Guarantees

Wireless Mesh Networks (WMNs) are going increasing attention as a flexible low-cost networking architecture to provide media Internet access over metropolitan areas to mobile clients requiring multimedia services. In WMNs, Mesh Routers (MRs) from the mesh backbone and accomplish the twofold task of traffic forwarding, as well as providing multimedia access to mobile Mesh Clients (MCs). Due to the intensive bandwidth-resource requested for supporting QoS-demanding multimedia services, performance of the current WMNs is mainly limited by spectrum-crowding and traffic-congestion, as only scarce spectrum-resources is currently licensed for the MCs' access. In principle, this problem could be mitigated by exploiting in a media-friendly (e.g., content-aware) way the context-aware capabilities offered by the Cognitive Radio (CR) paradigm. As integrated exploitation of both content and context-aware system's capabilities is at the basis of our proposed Active Mesh (ActMesh) networking paradigm. This last aims at defining a network-wide architecture for realizing media-friendly Cognitive Mesh nets (e.g., context aware Cognitive Mesh nets). Hence, main contribution of this work is four fold: 1. After introducing main functional blocks of our ActMesh architecture, suitable self-adaptive Belief Propagation and Soft Data Fusion algorithms are designed to provide context-awareness. This is done under both cooperative and noncooperative sensing frameworks. 2. The resulting network-wide resource management problem is modelled as a constrained stochastic Network Utility Maximization (NUM) problem, with the dual (contrasting) objective to maximize spectrum efficiency at the network level, while accounting for the perceived quality of the delivered media flows at the client level. 3. A fully distributed, scalable and self-adaptive implementation of the resulting Active Resource Manager (ARM) is deployed, that explicitly accounts for the energy limits of the battery powered MCs and the effects induced by both fading and client mobility. Due to informationally decentralized architecture of the ActMesh net, the complexity of (possibly, optimal) centralized solutions for resource management becomes prohibitive when number of MCs accessing ActMesh net grow. Furthermore, centralized resource management solutions could required large amounts of time to collect and process the required network information, which, in turn, induce delay that can be unacceptable for delay sensitive media applications, e.g., multimedia streaming. Hence, it is important to develop network-wide ARM policies that are both distributed and scalable by exploiting the radio MCs capabilities to sense, adapt and coordinate themselves. We validate our analytical models via simulation based numerical tests, that support actual effectiveness of the overall ActMesh paradigm, both in terms of objective and subjective performance metrics. In particular, the basic tradeoff among backbone traffic-vs-access traffic arising in the ActMesh net from the bandwidth-efficient opportunistic resource allocation policy pursued by the deployed ARM is numerically characterized. The standardization framework we inspire to is the emerging IEEE 802.16h one.