Efficient node cooperation stimulation mechanism in wireless multihop networks

In an autonomous environment of wireless multihop networks, like wireless ad hoc and sensor networks, nodes need to cooperate with one another to relay information effectively across the network. However, due to limited resources especially energy, nodes may be compelled to adopt selfish behaviour by not participating in forwarding packets for other nodes. Most, if not all, existing works on cooperation stimulation approaches assumed that this behaviour can be detected but do not explicitly describe how selfishness is actually quantified which leads to inaccurate behaviour judgment. This scenario may be worsened by false accusation issues induced by group of colluding nodes which might cheat in evaluating node behaviour during global/second-hand observation process, for their own communication gains in the network. In addition, cooperation of nodes is hampered by the lack of efficient scheme that is able to stimulate node cooperation effectively to achieve optimum communication rate in the network. This thesis proposes a mechanism named Compare and Measure Selfishness Detection (CMSD) to evaluate node behaviour more accurately and promptly by adopting quantification elements such as correlation co-efficient tool that are able to classify node into several selfishness/cooperativeness and fairness types. Based on simulation results, CMSD mechanism is able to provide more accurate and prompt behaviour information and increase network performance in term of low false positive and high packet delivery ratio. Moving further, this study proposes a mechanism that can avoid false recommendation and illegitimate node collusion during global observation. The mechanism is called Trust Features-based Evidence (TFE) which is developed based on significant and unique trust features to ensure that the reported information is of actual behaviour of an observed node. Assisted by an efficient cross-checking algorithm on several cheating and colluding cases, TFE mechanism is able to reduce the number of cheating nodes over time and the rate of illegitimate collusion as proven in the simulation results. Ultimately, a hybrid cooperation stimulation scheme named Recharge-As-Reward (RAR) and Credit-As-Reward (CAR) which functions as an explicit reward (i.e. in the form of token/flag) for cooperative nodes has been proposed in this study. Combining good features in reputation-based and credit-based mechanisms, this hybrid scheme is able to reduce the message/communication overheads suffered by many of existing hybrid schemes by reducing the reliance on a central agent. Performance evaluations demonstrate that the proposed mechanisms are able to reduce the number of dropped packets, increase nodes’ forwarding ratio and reduce message and communication overhead.

Preview Text FSKTM 2018 13 IR.pdf Download (2MB) | Preview

Actions (login required)

View Item