NeTS: Small: Control of Partially Observable Wireless Networks: Fundamental Limits Optimal Algorithms and Practical Implementation

1 Work to be Accomplished at Arizona State University The following work will be accomplished at Arizona State University. 1.1 Thrust 1: Partially Observable Wireless Access Networks We will start from wireless access networks where the access point is a central coordinator who collects network state information and then allocates resources accordingly. In this thrust, we will accomplish the following tasks. Task #1: We will consider partially observable access networks with general channel and traffic models and quantify the impact of limited and imperfect network state information. The fundamental limits, optimal algorithms and low complexity implementations will be developed. Task #2: We will characterize the delay constrained capacity of partially observable wireless access networks, and derive optimal algorithms to achieve the delay-constrained capacity. We will also develop algorithms with low-complexity and quantify the performance guarantees of low-complexity algorithms. 1.2 Thrust 2: Partially Observable Wireless Ad Hoc Networks In this thrust, we will consider wireless ad hoc networks, where no central coordinator is present and the algorithms need to be implemented in a distributed fashion. Control of partially observable wireless ad hoc networks is more challenging than that of wireless access networks because (i) it is very expensive to have every node/link to broadcast their states to all other nodes/links in the network and (ii) due to transmission latency, the network state information a node learns will differ from that of other nodes in the network, i.e., nodes will have inconsistent views of the network, which may lead to conflicting decisions and significantly reduce the network efficiency if the algorithms are not properly designed. The tasks to be accomplished in this project are: Task #3: We will characterize the delay constrained capacity of partially observable wireless ad hoc networks, and derive optimal algorithms to achieve the optimal capacity. We will also quantify the value of network state information in ad hoc wireless networks. Considering both the value of the information and the cost of disseminating it, we can determine to what extent the network state information should be made available. Task #4: We will develop distributed and low complexity algorithms for partially observable wireless ad hoc networks in which nodes have inconsistent network state information. Distributed algorithms inspired by statistical-physics will be derived, and the performance, in terms of both throughput and transmission latency, will be quantified. 1.3 Thrust 3: Evaluation and Implementation on the SoftNet Test Bed Task #5: We will implement some of the proposed algorithms and evaluate their performance on the SoftNet test bed at Arizona State University. The simulations and implementations will reveal important practical factors that should be considered but have been overlooked in network design, bringing in new theoretical challenges.