Mobile Beamforming & Spatially Controlled Relay Communications

摘要

We consider stochastic motion planning in single-source single-destinationrobotic relay networks, under a cooperative beamforming framework. Assumingthat the communication medium constitutes a spatiotemporal stochastic field, wepropose a 2-stage stochastic programming formulation of the problem ofspecifying the positions of the relays, such that the expected reciprocal oftheir total beamforming power is maximized. Stochastic decision making is madeon the basis of random causal CSI. Recognizing the intractability of theoriginal problem, we propose a lower bound relaxation, resulting to anontrivial optimization problem with respect to the relay locations, which isequivalent to a small set of simple, tractable subproblems. Our formulationresults in spatial controllers with a predictive character; at each time slot,the new relay positions should be such that the expected power reciprocal atthe next time slot is maximized. Quite interestingly, the optimal controlpolicy to the relaxed problem is purely selective; under a certain sense, onlythe best relay should move.