Abstract
Based on the ideas of conflict-free transmission, priority to guarantee transmission quality for communication between the different clusters, this article proposes a scheduling Algorithm fit for the MAC scheme of WSNs, which has made it possible for the polling service capable of differentiating services of the cluster head node of two priority levels. The high-priority service of the cluster head is responsible for routing between the different clusters, via exhaustive service policy, while the low-priority services of the cluster head node, for communication within the cluster through limited service policy with good fairness. The theoretical model of this scheme is established through Markov chain and probability generating function. Mathematical analysis is made on the mean queue length, the mean inquiry cyclic time and the mean delay time. It turns out that the findings from theoretical analysis correspond well with those from simulated experiments.
What Problem Does This Paper Address?
Wireless Sensor Networks (WSNs) consist of tiny, battery-powered nodes that must operate for long periods without maintenance. The most energy-intensive activity for these nodes is wireless transmission. In a hierarchical (clustered) network, Cluster Heads (CHs) have two jobs: collecting data from their local members and routing that data to other clusters or a central sink.
The challenge is that inter-cluster communication (routing) usually covers longer distances and requires more power and reliability than local data collection. Standard Medium Access Control (MAC) protocols often treat all traffic equally, which can lead to "bottlenecks" at the Cluster Head. If routing data gets stuck behind routine sensor readings, the entire network's latency increases, and energy is wasted on retransmissions or idle listening.
Methods
The authors propose a Routing Priority Scheduling Algorithm based on a polling mechanism. Polling is inherently conflict-free (no collisions), which saves energy compared to contention-based protocols like CSMA/CA.
The core of the method is a two-class priority system:
- High Priority (Key Node): This represents the routing interface between clusters. It uses an Exhaustive Service policy, meaning the Cluster Head will continue to transmit routing packets until that specific queue is completely empty.
- Low Priority (Common Nodes): These are the individual sensor nodes within the cluster. They use a Limited Service policy ($k=1$), where only one packet is served per polling cycle. This ensures that no single sensor node hogs the bandwidth, maintaining "fairness."
To validate this, the authors built a rigorous mathematical model using Markov Chains and Probability Generating Functions to calculate performance metrics like mean queue length and delay.
Key Insights and Results
- The "2N" Advantage: In this polling cycle, the "key node" (routing) effectively gets $2N$ times more service opportunities than common nodes. This ensures that inter-cluster data moves through the network at high speed.
- Stability and Efficiency: The authors define a clear stability threshold: the sum of traffic intensity ($\rho$) from all nodes must be less than 1. By using exhaustive service for routing, the "mean queue length" of the key node remains very small, reducing hardware requirements and packet loss.
- Validation: The findings from the theoretical analysis corresponded almost perfectly with simulated experiments (using 9 common nodes and Poisson traffic), proving the model's reliability.
- Performance: The mean delay for high-priority routing packets was significantly lower and more stable than for common nodes, directly contributing to longer battery life by reducing radio active time.
Limitations and Future Work
The mathematical model assumes symmetrical parameters and a static topology, which may not reflect "bursty" or mobile real-world WSNs. Future work could expand this to more than two priority levels to differentiate between emergency alerts and routine readings.
Summary updated