Cooperative estimation has been an active area of research in recent time. The idea underlying most state-of-the-art methods is to share information, obtained separately by subsystems comprising the system, and to use that information in an adequately designed (centralized or distributed) estimation algorithm that generates a common (global) estimate. The estimation performance that can thus be attained should be superior to that of each local estimator.
When the subsystems have only one global mission that needs to be accomplished (as opposed to local missions of the subsystems), one might think of better ways to exploit the existence of mutually cooperating subsystems. As the performance of each subsystem is not an issue, but, rather, the accomplishment of the global mission, each subsystem can relinquish the optimality of its own estimator, such that the global performance of the entire system would be optimized – at the cost of attaining sub-optimal performance by the local estimators. We call this cooperation strategy Altruism, taking after the well-known phenomenon exhibited in nature. Altruism has been used in sociology and in game theory to describe strategies that individuals would use in order to improve the chances of their species to strive.
In this seminar we will present two approaches for altruistic cooperation in the problem of parameter estimation by two cooperative estimators. In the first approach, termed heterarchical cooperative estimation, both estimators behave altruistically, sacrificing their own estimation performance for the purpose of improving global estimation performance. In the second approach, termed hierarchical cooperative estimation, one subsystem behaves egoistically (optimizing its own estimation performance) while the other behaves altruistically. When the underlying distribution is Gaussian, explicit results are obtained and demonstrated using a numerical example.
