A novel reputation computation model based on subjective logic for mobile ad hoc networks

Abstract

Mobile ad hoc networks are deployed under the assumption that participating nodes are willing to forward other nodes’ packets. However, for civilian applications where nodes are not owned by a single entity but are profit-oriented independent agents, cooperation cannot be taken for granted. In this paper, we present a novel reputation computation model to discover and prevent selfish behaviors by combining familiarity values with subjective opinions. The familiarity value represents a node’s familiar degree with another individual node. In our model, a node that queries another’s reputation first accumulates subjective opinions from their common neighbors. The familiarity values are used to calculate the weighting factor that determines how much a node’s recommending opinion impacts on the reputation computation result. The utilization of this familiarity allows nodes to obtain opinions with lower uncertainty values, which helps nodes to recognize selfish nodes much earlier and can decrease the convergence time for isolating selfish nodes. We evaluate the performance of our model based on ns-2 simulations to analyze the impact of different parameters on the network performance. The simulation results show that our model outperforms the pure subjective logic-based model and achieves up to a 25% improvement in the convergence time.

Introduction

Mobile ad-hoc networks have been the subject of intense research efforts for the last several years. Such networks consist of a set of mobile nodes that act both as terminals and routers, and do not rely on any infrastructure to communicate. A source node communicates with distant destinations using intermediate nodes as relays. Minimal configuration and quick deployment make ad hoc networks suitable for emergent situations, natural disasters or military conflicts, and it can also be used for temporary networks for conferences or expeditions. Since forwarding data for other nodes can drain the battery of a node, for civilian applications where nodes are not owned by a single entity but are profit-oriented independent agents, cooperation cannot be taken for granted. Users that want to maximize their own welfare and do not contribute to the network are defined as selfish nodes or free riders [1]. For the existence of such selfish nodes in the network, it is necessary to develop incentive mechanisms that avoid nodes that behave selfishly.

Currently, the incentive mechanisms can be divided into two categories: credit-exchange systems and reputation-based systems. In credit-exchange schemes [2], [3], [4], [5], [6], [7], [8], [9], nodes are awarded credits for forwarding messages, and these credits can later be used by these nodes to stimulate others to cooperate. The proposed systems that rely on the use of tamper-proof hardware to store credit information may hinder their ability to find wide-spread acceptance. While the system with a central control requires an infrastructure, which limits its wide application. In reputation-based schemes [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], a node’s behavior is measured by its neighbors using a watchdog mechanism [21], [22], and nodes can be punished for non-cooperation. However, detecting whether a given node actually forwarded a packet is not possible due to unreliable transmission in wireless networks. So cooperative nodes sometimes are perceived as being selfish, which causes a retaliation situation that can potentially decrease the throughput of cooperative nodes. To deal with this issue, Jøsang and Hayward [23] proposed a method based on subjective logic for discovering trust networks between specific parties. Subjective logic uses subjective opinions to express subjective beliefs about the truth of propositions with degrees of uncertainty and Kane and Browne [24] successfully transplanted and applied subjective logic to a wireless network environment. However, in wireless networks, a node’s behavior may be observed differently over time due to low connection reliability or nodes’ selfishness. Also a high uncertainty value leads to a longer convergence time for cooperative nodes isolating selfish nodes, which may encourage selfish behaviors since a long convergence time could make selfish nodes get as much of the benefit that could compensate them for their isolation.

To overcome the above deficiencies, this paper presents a novel reputation computation model which combines familiarity values with subjective opinions. The familiarity represents a node’s familiar degree with another individual node. In our model, a node that queries another’s reputation first accumulates subjective opinions from their common neighbors; then it combines the recommended opinions into one reputation opinion. The familiarity value is used to calculate the weighting factor that determines how much a node’s recommending opinion impacts on the reputation computation result. The utilization of this familiarity allows nodes to obtain opinions with lower uncertainty values, which helps nodes to recognize selfish nodes much earlier and can decrease the convergence time for isolating selfish nodes. We evaluate the performance of our model based on ns-2 simulations to analyze the impact of different parameters on the network performance, and we also compare our model with two other reputation models. The simulation results show that our model outperforms the pure subjective logic-based model and achieves up to a 25% improvement in the convergence time.

The rest of this paper is organized as follows. We first discuss some preliminaries in Section 2. We present our familiarity-based reputation model in Section 3. In Section 4, we demonstrate the simulation scenarios, for which Section 5 gives the analysis and provides the results. Section 6 presents the concluding remarks.