Abstract
With the increasing number of tasks undertaken by unmanned aerial vehicle (UAV) clusters, the corresponding flight ad hoc networks must process an increasing amount of service data, and different service types have different transmission requirements. According to these requirements, existing methods often prioritize high-priority packets at the expense of low-priority packets. To overcome this limitation, this paper proposes a dynamic time-division multiple access (TDMA) protocol based on service priority called the SPD-TDMA protocol. To meet the requirements of packet transmission with different priorities, a queue-scheduling algorithm based on maximum waiting time was designed. In this algorithm, packets received from upper-layer applications are allocated to different queues according to their priority, and a corresponding waiting time step is allocated to each queue. Packets with higher priority have shorter waiting times. When the transmission time slot arrives, the packets with the shortest remaining waiting time are sent preferentially, which not only meets the transmission requirements of high-priority packets but also effectively avoids the starvation of low-priority packets. On this basis, to optimize the usage of time-slot resources further, the SPD-TDMA protocol adopts a priority predictive time-slot scheduling mechanism to avoid overbooking time slots or wasting idle time-slot resources to improve the overall efficiency and throughput of the network. Simulation results based on OMNeT++ indicated that compared with the IEEE 802.11DCF, TDMA, and P-TDMA protocols, the proposed method increases the packet delivery rate by 80%, 58%, and 27%, and throughput by 85%, 58%, and 29%, respectively.
Introduction
In Flying Ad Hoc Networks (FANETs), UAV clusters must share limited wireless channels under conditions of high mobility and dynamic topologies. Traditional protocols face significant hurdles: competitive methods like CSMA/CA (IEEE 802.11DCF) suffer from collisions in dense networks, while static TDMA often results in inefficient resource utilization. Moreover, standard priority-based systems frequently lead to "starvation," where low-priority data is indefinitely delayed by high-priority traffic. The SPD-TDMA protocol addresses these challenges by integrating a fair priority queuing mechanism with predictive time-slot scheduling.
Methodology
SPD-TDMA introduces two core innovations to manage network traffic efficiently. First, a Maximum Waiting Time Queue Scheduling algorithm prevents starvation by assigning a "waiting time step" to each packet. As time progresses, the urgency of low-priority packets increases, ensuring they are eventually transmitted. Second, Quadratic Exponential Smoothing Prediction allows nodes to estimate their future time-slot requirements based on historical demand, enabling dynamic resource allocation.
The protocol's operation is structured into three phases per frame: the Claim Phase for resource estimation, the Response Phase for conflict resolution based on priority weights, and the Info Phase for actual data transmission.
Results and Discussion
The protocol was evaluated using the OMNeT++ simulator with 40 to 80 UAVs moving according to a Gauss-Markov model at speeds between 72 and 216 km/h. Compared to IEEE 802.11DCF, standard TDMA, and P-TDMA, the SPD-TDMA protocol demonstrated superior performance across all key metrics. It achieved an 80% increase in packet delivery rate (PDR) and an 85% improvement in throughput over 802.11DCF.
As shown in the performance analysis, while 802.11DCF performs well under light traffic, SPD-TDMA maintains significantly lower and more stable end-to-end delays as network load and mobility increase, offering a 55% delay reduction compared to standard TDMA.
Conclusion
The SPD-TDMA protocol effectively balances the need for high-priority data transmission with the requirement for overall network fairness in FANETs. By utilizing predictive scheduling and a time-sensitive queuing model, it significantly enhances throughput and reliability in high-mobility scenarios. While the protocol introduces higher computational complexity ($O(n^4)$) for conflict resolution, its performance gains are substantial. Future research will explore the integration of network coding and more advanced heuristic algorithms to further refine scheduling efficiency.
Summary updated