In recent years, device-to-device (D2D) communication has attained significant attention in the research community which enables the users to exchange data directly with each other without the intervention of Base Station (BS). The advantages of D2D communication can be fully realized in multi-hop communication scenario. Establishing a multi-hop D2D route is crucial whenever the two nodes which want to communicate are not in the transmission range of each other. Routing in such multi-hop cellular D2D networks is a critical issue, since the multi-hop network can perform worse than a traditional cellular network if wrong routing decisions are made. This is because routing in these multi-hop networks needs to consider the challenges of node mobility, dynamic network topology, and network fragmentation, which are not major concerns for traditional cellular network.
This thesis mainly focuses on routing in multi-hop D2D networks and provides three main contributions as follows:
In first contribution, firstly, a discussion of various implementations, performance improvements and incentive mechanisms to support D2D communications is provided. Secondly, a detailed derivation and explanation of taxonomy of state-of-the-art routing algorithms for D2D networks is provided. It is worth mentioning that there are numerous works over the past two decades on routing for D2D networks with different naming conventions other than D2D communications. Thus, to the best of the knowledge, the provided taxonomy encompasses all possible routing algorithms. Each routing scheme consists of two main parts: routing metric, and route discovery mechanism. This work contributes in both parts by proposing a routing metric, MIIS, in second contribution and reactive and proactive centralize route discovery mechanisms in third contribution.
In second contribution, a novel routing metric, MIIS (Metric for Interference Impact and SINR) is proposed. Interference is an important parameter to consider in D2D communication, therefore, interference together with signal-to-interference-plus-noise-ratio (SINR) are considered in the proposed routing metric, MIIS. The MIIS selects routes having higher SINR while being conscious about lower interference at the same time. The performance of MIIS is evaluated in a distributed D2D network implemented in OMNeT++ network simulator. Different MIIS variants are compared with two state-of-the-art schemes in terms of average hop count, routing overhead, packet loss ratio and end-to-end delay. The simulation results showed that all MIIS variants outperform the state-of-the-art schemes under various network topologies with varying number of nodes, nodes mobility and traffic load.
In third contribution, a novel route discovery mechanism of reactive centralized routing is proposed that takes the advantage of the presence of BS for establishing the D2D routes in which the route calculation and decision are taken by the BS for a given cell. The BS makes route decisions by gathering the information from all the nodes and constructs the network topology graph. The reactive centralized routing is also extended to proactive centralized routing. Both centralized routing schemes (i.e., reactive and proactive) significantly reduce the routing overhead as compared to distributed routing schemes by avoiding the flooding of route requests. The performance of reactive and proactive centralized routing have been evaluated through simulations by using MIIS and comparing with existing schemes in terms of varying number of nodes, nodes mobility and traffic load. The centralized routing schemes achieves much lower routing overhead, packet loss ratio and end-to-end delay.