A Mutual Compensation Mechanism for Efficient Decentralized Network Management: The Case of Power Control in Wireless Networks

A “mechanism ” is a set of rules governing the interaction of selfish entities, which attempts to lead these entities to a desirable outcome. This work applies a relatively simple mechanism, available in the economics literature, to achieve an efficient decentralized allocation of power among data-transmitting terminals. The resulting operating point is “efficient”, because terminals end up “fairly ” compensating each other for the interference each one causes. The same ideas can be fruitfully applied in more general networks, and even outside the engineering context. It has long been recognized that decentralized control algorithms offer many advantages over their centralized counterparts. Reasons include complexity, signaling overhead, and unavailability of local information to a central controller. Besides, cer-tain modern communication and/or computing paradigms, such as ah-hoc wireless networks, and peer-to-peer computing, are inherently decentralized, which make central controllers highly impractical, if not outright impossible to implement. With minimal or no intervention by a central authority, many economic agents, each acting independently, in its own interests, and making relatively simple decisions, can achieve sensible outcomes. Thus, a free economic market provides the engineer with a useful paradigm for decentralized control, and microeconomics and game theory can provide a solid scientific foundation. For instance, power control in wireless data applications has been formulated as a “game”; i.e., a situation in which each of several selfish agents choose a “strategy ” in order to maximize its own “payoff”, which depends on the chosen strategie