Packet Scheduling Algorithm Based on Priority Adjustment in Wireless Sensor Networks

The quality of service in wireless sensor networks is significantly affected by delays. The existing dynamic multilevel priority (DMP) scheduling algorithm suffers from the problem of starvation where low-priority services cannot be fulfilled in time, leading to higher group delay in the case of sudden traffic flow. To address the limitations of the DMP algorithm, this paper proposes a priority dynamic adjustment packet scheduling (PAS) algorithm to effectively reduce the queuing delay of packets in a node, improve the starvation problem associated with low-priority services, and ensure the quality of different services. The algorithm optimizes the queuing method of different service types in the ready queue and is dynamically adjusted according to the service type and backlog of packets. Consequently, queuing delays of non-real-time service packets are reduced, thereby leading to a reduction in end-to-end delays. The simulation results show that, in a scenario with a high arrival rate of service flows, the PAS algorithm can reduce the end-to-end delays by 30% compared with the DMP scheduling algorithm.

I. Introduction

Owing to extensive application of wireless sensor networks (WSNs), many scenarios require a guaranteed quality of service (QoS) for data flows in the network. These scenarios include factory production line monitoring systems, medical service systems, and smoke alarm systems. Owing to the limited bandwidth and many-to-one data flow characteristics of WSNs, providing differentiated services is challenging. The data in WSNs can be divided into real-time and non-real-time data according to business types. For example, in an industrial production line monitoring system, alarms for an abnormal operating state of the equipment are time-sensitive. The data from such alarms are called real-time data. Real-time data must be transmitted to the destination node as soon as possible to ensure that the delay requirements are satisfied. In multi-hop forwarding, packets in WSNs need to be queued within nodes. The queue waiting time on multiple nodes is an important factor that affects the end-to-end delay of packets. Therefore, a scheduling algorithm is required to schedule the packet forwarding process.