Introduction - Social Group Utility Maximization - page 2

relationships with other users are generally heterogeneous , it would take into account
the effect of its behavior on different users to different extents, which is not captured
in the existing studies.

1.2 Cooperative Wireless Networking

Node cooperation has been widely recognized as a promising strategy for a variety
of wireless networks. Indeed, individual nodes can achieve significant performance
gain by cooperating in a coordinated way. For example, cooperative communica-
tion is an effective approach for improving the transmission rates among nodes in
a communication network. In a cognitive radio network, cooperative sensing can
enable cognitive radio (CR) users to efficiently detect spectrum opportunities that
are not used by primary users (PUs). Although the benefit of node cooperation is
pronounced, cooperative behaviors come at the cost of the cooperative nodes (e.g.,
in terms of the resource consumption devoted to cooperation). Therefore, for a net-
work consisting of autonomous users, users may not be willing to cooperate without
adequate incentives.
There exist numerous studies on incentive design for stimulating user coopera-
tion for networking. Existing work on this subject can be broadly classified into three
categories. One category of work makes use of reciprocity (also known as barter )
[ 8 - 11 ]. Although a reciprocity-based approach is simple to implement, it is inefficient
in general since it is rare to have synchronously matched requests for cooperation.
Another category is based on (virtual) currency [ 12 - 15 ], in which a user earns cur-
rency by providing service to others and spends currency to receive service from
others. The use of currency as a medium of exchange overcomes the shortcoming
of reciprocity-based approaches by enabling users to “asynchronously trade” coop-
eration. However, a major drawback of using currency is that it incurs a significant
implementation overhead, mainly due to the need to inhibit malicious manipulation
among users without mutual trust. Consider, for example, the Bitcoin [ 16 ] that has
recently drawn widespread attention as a digital currency. The creation and transfer
of bitcoins need to consume considerable computing resources so that they can be
secured against potential cheating using cryptographic tools. Reputation-based ap-
proaches [ 17 - 19 ] constitute the third category. Since reputation score can be viewed
as a form of currency, these approaches share the same advantages and disadvantages
as the currency-based ones.
For a network consisting of autonomous users (nodes) (e.g., ad hoc networks),
each user may act in a selfish manner, in the sense that it only cares about its own
benefit (e.g., utility) and does not care about the effect of its behavior on other
users. In this case, the strategic interactions among users can be modeled by a
non-cooperative game (NCG), where each user aims to maximize its payoff. NCG
has been extensively studied for wireless networking applications [ 20 ]. Due to the
selfish nature of users, the stable outcome of a non-cooperative game (e.g., a Nash
equilibrium) may achieve a low social welfare (i.e., the total benefit of all users).