Packet Delay in UAV Wireless Networks Under Non-saturated Traffic and Channel Fading Conditions

In this paper, we conduct a statistical analysis for the packet delay in a wireless network of unmanned aerial vehicles (UAVs) under non-saturated traffic and channel fading conditions. Each UAV runs the distributed coordination function of IEEE 802.11 at the medium access control layer, and all UAVs are one-hop neighbors. A pair of UAVs can communicate over the lossy wireless channel of a fixed data rate. A non-saturated traffic condition is used. By modeling each node UAV as a standard queueing system (i.e., $M/M/1$ or $M/G/1$ queue), we derive the mean packet delay under the non-saturated traffic condition. Numerical and simulation results show that the mean packet delay derived based on $M/M/1$ queue is accurate for UAV wireless networks under the non-saturated traffic condition and with an independent packet error rate. It is observed that the mean packet delay increases with either the number of UAVs in the network or the packet generation rate. More important, existing results in the literature, based on the saturated traffic condition (i.e., packets are always supplied for transmission), tend to overestimate by a large amount the mean packet delay for networks with non-saturated traffic. In the second part of this paper, we apply simulation data to analysis of the probability distribution function of the packet delay when the packet error rate equals zero. Using a distribution fitting tool, we observe that the packet delay can be well approximated by the sum of a deterministic delay, which corresponds to the time period during which the UAV senses the medium and is able to perform a successful transmission, and a random delay, which follows a Gamma distribution function.