Incentives for Resource Sharing in Ad Hoc Networks: Going Beyond Rationality

Introduction

For over a decade, mobile ad hoc networks (or MANETs) have been in the centre of attention of the networking research community. They have created significant anticipation for the new step in the telecommunications history: It is expected that people will be transformed to “smartmobs” that will self-organize in creating their own ad hoc communications networks to exchange all types of information, provide access to Internet gateways, socialize, and participate themselves or through their device in a large variety of peer-to-peer applications like file sharing and ubiquitous computing.

Yet until now this vision has not become a reality although significant progress has been made over the last years regarding the wireless technology (bluetooth, 802.11), the capabilities and resources of the portable devices, and the required protocols for routing, media access, and power control. Maybe a few more steps are required in order to reach the point to enable an interesting set of applications on top of a general multi-hop ad hoc network. But people are most probably the critical piece that is still missing: Users are the key component of a mobile ad hoc network because they are both the providers and the consumers of the service (and this is so at all layers of the system’s architecture). In other words, MANETs are peer-to-peer (p2p) systems whose successful operation depends highly on the users’ desire to participate and contribute the resources of their devices.

Participation and resource sharing are the main requirements for all p2p systems to bootstrap and survive. Significant part of ongoing research is devoted on the design of incentive mechanisms for resource sharing. However, in the case of mobile ad hoc networks this problem becomes more challenging due to the critical mass that is required for the creation of an operational ad-hoc network, the short time scales in which users might reside in the same place, and the often scarce resources that should be contributed (i.e. battery, processing power, and bandwidth). Additionally, more research is required in terms of application design to encourage participation, since the “killer application” seems to be still missing. Finally, one should take into account also psychological factors that are related to the fact that people are not used to this direct form of communication with strangers.

The most fundamental resource sharing issue in this context is packet forwarding. There is a wide variety of economic mechanisms proposed in the literature. They either assume the existence of a virtual or real currency management system or they follow a reciprocity-based approach that adopts the notion of reputation in accounting for a user’s past behavior (and provide rewards or punishment accordingly). In addition to packet forwarding, however, there are many more levels of cooperation in an ad hoc network. So, there is also work on incentive compatible topology formation and energy efficient routing, and incentives for cooperation at the lower layers of the network architecture (media access, and power control).

In general, there are two different dimensions in the problem formulation: the economic and the technological. The economic dimension is related to the conceptual mechanisms, the high-level rules of the system that are required to decide on certain levels of consumption and/or contribution for different users based on certain assumptions their utility, cost, and behavior (the economic model). The technological dimension is related to the technical means for enforcing the chosen incentive mechanism. That is, the mechanisms required to monitor and assess the behavior of users, to account for it over time, and implement the rewards and punishments prescribed by the incentive mechanisms according to each individual’s behavior. Unfortunately, the constraints that are posed by technology in terms of enforcement are often the “bottleneck” in this environment, and the choices for the potential high-level incentive mechanisms become restricted.