Stochastic Optimal Control Algorithms and Next Generation Technologies for Dynamic Resource Allocation in Mobile Communications Networks

The overall objective of the proposed Phase II effort is the development and evaluation of the OX-MAC (Optimized NeXt Generation MAC) suite of protocols to provide optimal media access control for mobile wireless networks formed by devices capable of dynamic spectrum allocation. OX-MAC is composed of XM-OptimalKernel, which disseminates transmission opportunities across the network, and XM-OptimalAccess, which allocates these opportunities on an optimal fashion. Scheduling algorithms in XM-OptimalAccess will be developed on the basis of Stochastic Optimal Control and Graph Theoretic techniques. Phase I has demonstrated the utility of these techniques for designing schedulers for CDMA2000 high-data rate systems. Specifically, in Phase I we designed a scheduler taking into account dynamic spectrum allocation, and demonstrated substantial increase in performance relative to current systems. Phase II will utilize these tools for designing OX-MAC for three classes of networks: (1) cellular type networks with a single base station, (2) cellular type networks with multiple base stations, and (3) Mobile Ad-Hoc Networks. OX-MAC will be coded and evaluated using high-fidelity OPNET platforms. BBN Technologies, a key contractor on DARPA's Next Generation Communications program, will provide unique expertise in dynamic spectrum allocation, as well as software modules to support the implementation and evaluation of OX-MAC. BBN will also provide commercialization support. Prof. Harold Kushner from Brown University and Prof. Sanjoy Mitter from MIT will be consultants in Stochastic Optimal Control, Queuing Theory, and Information Theory.