Data transmission in multi-hop wireless network under uncertainty: Prospect theory approach

Wireless network services are intended to play a role of primary importance in the immediate future and the traditional centralized, fixed networks may no longer satisfy the growth in demand for making wireless connections faster in an ever-increasing mobility. New concepts related to infrastructure-less networks, therefore, are attracting a lot of attention nowadays. The reason is that they allow to get a more scalable and flexible wireless networking, along with an efficient power usage, when fixed network infrastructures are not available. In these networks, nodes are guided by forward-looking goals and they are able to converge on-the-fly to arbitrary form graphs that allow them to provide wireless services with¬out requiring a centralized control entity. An interesting problem arises from the fact that the participants do not necessarily have an incentive to cooperate with each other. Because of that, the selfish behavior of each node in the network has made it very difficult to find a solution with traditional mathematical analysis. In order to deal with such a problem, the idea of modeling the topology from a game theoretical perspective has become particularly attractive. This is because game theory provides the tools to model individual, independent decision-makers, whose actions potentially affect all other par¬ticipants. Thus, it resulted to be remarkably effective in selecting the best node of the network to connect to and thereby achieving excellent performance. To understand the implications of these issues, a game theoretical framework aimed at minimizing the transmit power in ad-hoc wireless network has been designed. A common message, e.g. a video with a specific quality and the same length, in a video streaming scenario, has to be sent from a fixed source to all the nodes of the network in a multi-hop manner. We wanted to address three main issues that arise in the considered network scenario. One concerns the fact that nodes can join and leave the system at any time, which means that the connection conditions can continuously change over time. The second is that in a multi-hop network an incentive is very important for intermediate nodes, because if they have to relay a large amount of information, a pricing mechanism may be required for forwarding and cooperation. As last, many related works consider all nodes to be empty (did not download any data) when the system in analyzed. On the contrary, in reality, some nodes could start receiving and at the same time other devices could join the network. To model the preference with which a node decides the service provider from which to receive the message, two models have been considered: Expected Utility Theory (EUT) and Prospect Theory (PT). The first led us in having always rational choices, but totally uninfluenced by real-life perceptions. The second, instead, takes into account the fact that end-users may interfere with the decision made by the underlying system design. From the simulation results we show that a deviation from the behaviors expected from the application of the EUT can lead to a degradation of the total energy spent in the network. However, in a dynamic environment, recovery from a failure can definitely require additional resources and according to the rules of the PT this circumstance is successfully avoided. Based on these two modes we will see how the network should be formed when the information is transferred within a limited contact duration.