Decentralized Monitoring in Mobile Ad Hoc Networks - Provisioning of Accurate and Location-Aware Monitoring Information

There has been an immense popularity increase of mobile communication devices, such as smartphones and tablet PCs, due to the technical progress in equipment manufacturing and communication. Paired with the constantly growing availability of wireless broadband access over cellular networks the devices enable the continuous consumption of applications. Furthermore, they foster new types of applications, such as location-based services or mobile social networks, where the user interacts with its vicinity or nearby users. Due to the increasing popularity paired with the continuous consumption of applications cellular networks can hardly handle the resulting traffic that frequently exceeds their capacity. To unburden these networks mobile ad hoc networks represent a viable alternative, complementing or superseding cellular networks. Exploiting the growing density of mobile communication devices, mobile ad hoc networks rely on the direct interconnection of devices and provide an useful substrate to exchange information and to deploy applications. In particular, the direct device interaction reflects the communication and interaction pattern of location-based services and mobile social networks, which we term as locality of interaction. Despite the fact that mobile ad hoc networks represent a viable communication substrate, they are exhibiting highly dynamic characteristics, which are mainly attributable to the autonomous behavior of users. To handle the dynamic nature mobile ad hoc networks must be adapted to the current state of the network and the influencing conditions of the surrounding environment. In this regard, monitoring constitutes an inevitable aspect of mobile ad hoc networks. It collects information from the users and provides essential insights into the current network state, serving as basis to adapt the network. As mobile ad hoc networks are exclusively established by users without a central entity the users themselves must monitor the network and exploit the obtained monitoring information to adapt the network. Consequently, the issues arise that the users are in charge to measure and collect the monitoring information as well as to distribute the obtained insights among them. In addition to the information about the network state a user requires detailed information about its vicinity. The reason for the provisioning of location-dependent information results from the locality of interaction in mobile ad hoc networks: as a user can only interact with its neighborhood it particularly requires detailed monitoring information about its vicinity. In the context of monitoring we refer to this property as location-aware monitoring. To address the presented issues decentralized monitoring mechanisms have been introduced and developed. The participating users accomplish the mentioned tasks of monitoring and provide the required monitoring information. However, existing approaches for decentralized monitoring in mobile ad hoc networks exhibit certain shortcomings or make some limiting assumptions. Examples comprise (i) the need for a decentralized infrastructure with dedicated devices that monitor the network or (ii) the expectation of an additional external entity (e.g., a network operator, which further processes the monitoring information). The examples reveal that existing approaches do not comply with the addressed issues and are not suited for the deployment in the envisaged scenarios. Consequently, the major objective of this thesis is the design of decentralized monitoring approaches that tackle these issues and are applicable in the corresponding scenarios. BlockTree.KOM represents our first solution for decentralized monitoring in mobile ad hoc networks. It proposes a set of four fundamental concepts that specify its underlying structure and the exchange of information over that structure. In detail, BlockTree.KOM relies on a hierarchical topology and on its own tailored communication methods to monitor the communication network. With P-BlockTree.KOM and C-BlockTree.KOM we present two different approaches of BlockTree.KOM that comply with the presented concepts but implement them in different ways. Based on an extensive evaluation, the obtained results reveal that both approaches provide accurate monitoring information despite the dynamic nature of mobile ad hoc networks. Exploiting the hierarchical topology, specifically P-BlockTree.KOM is able to provide location-aware monitoring information: every user obtains an accurate view of its vicinity as well as summarized insights on distant regions. Mobi-G.KOM constitutes our second solution for decentralized monitoring in mobile ad hoc networks. Its design fundamentally differs from BlockTree.KOM, since it operates on a flat topology and abstains from the establishment of a hierarchy. To implement the communication between users over the flat topology Mobi-G.KOM exploits the robust communication pattern of gossiping. With the flat design of Mobi-G.KOM we sacrifice the provisioning of location-aware monitoring information. The obtained results reveal that the view of a user on nearby or distant regions does not differ and leads to inaccurate information about a user’s vicinity. However, the experiments unveil the advantages of the flat design paired with the robust communication pattern of gossiping. Mobi-G.KOM is highly robust and operates in highly dynamic environments. Furthermore, the provisioning of accurate monitoring information comes at considerably lower cost: Mobi-G.KOM neither produces traffic to maintain a hierarchical topology nor transmits data at multiple levels of the topology