Optimal Semistable Control in Ad Hoc Network Systems: A Sequential Two-Stage Approach

Abstract

Motivated by various problems such as distributed computation, multiagent coordination, wireless communication, and online search algorithms, a time-varying optimal coordinated information load balancing problem has been solved by means of a sequential, two-stage, optimal semistable control approach. Technically we formulate this information load balancing problem into a linear, time-varying quadratic semistabilization problem with time-dependent iterative algorithms for information load balancing in peer-to-peer networks. To solve this problem, we propose a novel, sequential two-stage design. The first stage is to guarantee the convergence of the optimal policy while the second stage is to derive the explicit recursive formulas for optimal strategies under a finite set of convergence-guaranteed candidate policies.