QoS-aware Relaying in Ultra-low Latency Wireless Communication

Abstract

Wireless networking is not yet able to meet the strict requirements of the industrial world. Despite offering a bunch of advantages, it struggles because of the harsh requirements to latency and reliability posed by some of the applications. In this thesis, a method is presented that enables the differentiation between traffic classes on the MAC layer. This allows the reallocating of resources to prioritize the traffic with higher requirements. Therefore, each node in the network is buffering overheard traffic and dynamically takes over the role as relay. By letting each node decide on its own whether it should act as a relay or not and using TDMA to control the medium access, the need for short-noticed communication between the different entities in the network can be eliminated. An experimental evaluation indicates that the presented protocol design achieves a clear prioritization, without reducing the QoS provided to the lower priority traffic. Reliabilities of up to ${10}^{-6}$ could be measured for a subset of traffic, despite raw packet error rates of over ${10}^{-2}$. Even if no priority classes exist in a network, the proposed protocol increases the reliability by over an order of magnitude. Additionally, the reliability is scaling with the utilized capacity of the network, something not seen by typical TDMA solutions. The presented solution is able to offer low latency and high reliability at the same time, but not at the levels necessary for some of the industrial applications. Nevertheless, an improvement can be seen and some potential future work is presented which is expected to move the protocol’s performance ever so slightly closer to achieving this goal.