Experimental Evaluation of Statistical Delay Bound for IEEE 802.15.4 TSCH Networks
The adoption of wireless sensor networks (WSN) in industrial applications has raised in the last decade. Networks based on the IEEE 802.15.4 standard TSCH (time slotted channel hopping) medium access have been selected for these applications, thanks to low power and reliable communication. Quality of service (QoS) for these applications must be provided in terms of end-to-end (E2E) delay and reliability. In this work, we carry out the QoS analysis of these networks applying stochastic network calculus (SNC). We setup a WSN to study the stochastic network performance of the wireless channel, and model it using the SNR distribution and its relationship with bit transmission errors. Channel model is used to derive the service curve of the network, which accounts for random properties of the channel and communication schedule used in the network. An expression for an E2E upper delay bound on delay violation probability (DVP) is derived for the steady state network. The DVP is the probability that packets arrive later than expected target delay, which indirectly expresses the reliability requirement of the network for a certain target delay. We have evaluated the bound using simulations and an experimental setup. The experimental evaluation is performed in a controlled environment imitating the channel conditions observed in real environments. We compare and discuss the dierentexperimental results for increased arrival rates and various channel conditions. The results show that the computed bound is an upper bound on the DVPs from simulation and empirical setup. However, the bound is gets looser with good channel conditions. By observing numerous empirical and simulation results, we discover that the analytical bound is overestimating the DVP by one target delay, this behavior of the bound needs to be further investigated in the future work. We found that experimental results match well with simulation results.
München, TU, Master Thesis, 2019