Dealing with Selfishness and Moral Hazard in Noncooperative Wireless Networks

For noncooperative networks in which each node is a selfish agent, certain incentives must be given to intermediate nodes to let them forward the data for others. What makes the scenario worse is that, in a multihop noncooperative network, the end points can only observe whether or not the end-to-end transaction was successful or not, but not the individual actions of intermediate nodes. Thus, in the absence of properly designed incentive schemes, rational and selfish intermediate nodes may choose to forward data packets at a very low priority or simply drop the packets, and they could put the blame on the unreliable channel. In this paper, assuming the receiver is a trusted authority, we propose several methods that discourage the hidden actions under hidden information in multihop noncooperative networks with high probability. We design several algorithmic mechanisms for a number of routing scenarios such that each selfish agent will maximize its expected utility (i.e., profit) when it truthfully declares its type (i.e., cost and its actions) and it truthfully follows its declared actions. Our simulations show that the payments by our mechanisms are only slightly larger than the actual cost incurred by all intermediate nodes.