Achieving Pareto-Efficient Bandwidth Allocations Using a Non-Monetary Mechanism

In this paper, we present a mechanism on a non-monetary level that achieves Pareto-efficient bandwidth allocation in a network. Monetary mechanisms, known as congestion pricing, are subject to implementation issues and opposed by conceptional questions in a scenario on end user level. Therefore we conducted our research in the field of non-monetary bandwidth allocation mechanisms. Non-monetary mechanisms suffer from the problem of preference elicitation. The issue of preference elicitation is predominantly solved using implementations of the Clark-Groves mechanism, such as a Vickrey auction. At present there exists no applicable transformation of a Clark-Groves mechanism on a non-monetary level. Therefore we propose a mechanism that alters users strategy space via incentives such that truthful preference elicitation becomes a dominant strategy in a repeated game. Our mechanism is based on the intuition that preferences can be elicidated via implied behaviour. It is thus possible to obtain Pareto-efficient allocations if users get incentives to only ask for bandwidth if their marginal utility exceeds marginal social costs. Furthermore, the resulting bandwidth allocation possesses the desired property of Nash bargaining fairness