Integrated Beam Steering and Scheduling for Spatial Reuse

Increasing spatial reuse of spectrum is fundamental to improving the capacity and usefulness of large wireless networks. Steerable and adaptive antennas facilitate reuse, and a number of MAC-layer protocols have been developed for them. Existing protocols, however, do not allow the MAC (or scheduling) process to fully consider the capabilities of antenna reconfiguration, or vice-versa. Without such integration between MAC scheduling and physical antenna configuration, a “chickenand-egg” problem exists: If antenna decisions are made before scheduling, they cannot be optimized for the communication that will actually occur. If the scheduling decisions are made first, the scheduler cannot know what the actual interference and signal strength properties of the network will be. To take full advantage of the dynamic capabilities of an advanced antenna system, MAC-layer processes need to know about not just any particular RF-level configuration, but about the range of possible configurations. This research has two goals: To produce a fully-integrated algorithm for STDMA scheduling and beam steering, and to systematically evaluate the costs and benefits of different approaches to such integration. Our approach to this research is based on a combination of mathematical analysis, simulation, implementation and deployment, and empirical measurement. I. RESEARCH OBJECTIVES We evaluate the potential gains from a tighter integration between MAC-layer scheduling and advanced antenna system configuration. There is substantial existing research on MAC layers for smart antennas, but none of it gives significant attention how the MAC and antenna configuration systems interact. When the MAC-layer decision process takes the antenna configuration as given, it is unable identify situations in which a different antenna configuration would create better options at the MAC layer. Conversely, if the MAC layer assumes that a given set of concurrent transmissions can be accommodated by physical-layer antenna reconfiguration, it runs the risk of creating a high-interference situation in which none of the links function well. That risk can be mitigated by conservative assumptions, but that means passing up legitimate opportunities for spatial re-use. This research is focused specifically on spatial-reuse TDMA, both because it has promising performance in its own right, and because an explicit scheduling process is easier to evaluate and improve than the emergent scheduling behavior of a contention-based MAC. STDMA tends to perform better than CSMA in terms of spatial re-use, and STDMA with smart antennas performs better than with omni-directional ones [1]. Even so, it is easy to construct artificial scenarios in which the best current smart-antenna STDMA algorithms achieve less than half the performance of an integrated algorithm. It remains to be seen how significant the gains from integration will be in “naturally-occurring” scenarios, but that is one of the questions being addressed. We are developing and evaluating specific integrated algorithms for beam selection and scheduling, and on systematically evaluating the costs and benefits of such integration in general. It is expected that this research will provide both useful prototype systems and general guidance for using smart antennas in dense wireless networks.