Research On Trust And Energy-Efficient Clustering Algorithms In Ad Hoc Networks

In the recent few years, contemporary techniques on wireless communication, embedded computing and networking evolve very rapidly. Consequently, numerous emerging applications, including military battle field communication, disaster and emergency treatment, evolves with increasing demands for connecting mobile nodes to build up a temporary network without any underlying fixed infrastructures. Ad Hoc wireless networks are then proposed for fulfilling these requirements, and thus quickly become a hot research topic.However, comparing with traditional networks, Ad Hoc networks have several distinguished characteristics, such as decentralized nature, self-organization, multi-hop communications, and so on. These features, thereafter, bring great challenges to several networking issues, including scalability, route maintenance, and network management. To handle with these challenges, the hierarchical clustering architecture is proposed, but building such architecture greatly depends on its fundamental algorithmic aspects, i.e., the clustering algorithms. To date, existing clustering algorithms mainly focus on several generic technologies, including energy-saving, load balancing, handling external security threats, etc. Nevertheless, they fail in providing both the resistance capability to interior malicious network attacks and the energy optimization approaches for emerging applications. Currently, trust has been proven to provide effective means towards the interior malicious network attacks, while energy efficient clustering techniques should be built upon a comprehensive analysis with specific network applications. Therefore, there are urgent requirements to developing new clustering algorithms in order to overcome these two drawbacks.This thesis focuses on designing trust and energy efficient clustering algorithms in Ad Hoc networks. In general, our contribution can be summarized as follows:Firstly, towards the serious trust threat among Ad Hoc networks, SACA (Spreading Activation based Clustering Algorithm), an innovative trust evaluation clustering algorithm is proposed. SACA introduces the spreading activation model into the trust evaluation clustering algorithms in Ad Hoc networks, and penetrate the entire process of cluster head election, joins of cluster members and cluster maintenance. Experimental results show that: the SACA algorithm builds an effective mechanism for evaluating the trust among Ad Hoc networks, and provides an effective measure towards the interior malicious network attacks.Secondly, to meet the requirements of energy efficient algorithms in emerging Ad Hoc sensor networks with period collection, a novel energy efficient clustering algorithm, named ISCA(Information redundancy Suppression-based energy efficient Clustering Algorithm), is proposed based on the idea of redundancy information depression. Considering the distinguished characteristics of this kind of sensor Ad Hoc networks, ISCA takes full advantage of the temporal and locality redundancy properties of sensing data, and then incorporates a careful design with the cluster head election and cluster forming process. In the cluster forming stage, ISCA works by introducing the concept of data collection credit, and combines this metric with the residual energy information into the cluster head election process, and then build clusters dynamically. In the data transmission stage, ISCA performs both temporal and spatial data suppression with redundancy information in order to reduce unnecessary data transmissions. Experimental and analysis results show that: comparing with existing algorithms, ISCA can not only prolong the network lifetime substantially by reducing network energy consumption effectively, but also shows its scalability with different network sizes and different deployment densities.