A Fair and Efficient Packet Scheduling Scheme for IEEE 802.16 Broadband Wireless Access Systems

This paper proposes a fair and efficient QoS scheduling scheme for IEEE 802.16 BWA systems that satisfies both throughput and delay guarantee to various real and non-real time applications. The proposed QoS scheduling scheme is compared with an existing QoS scheduling scheme proposed in literature in recent past. Simulation results show that the proposed scheduling scheme can provide a tight QoS guarantee in terms of delay, delay violation rate and throughput for all types of traffic as defined in the WiMAX standard, thereby maintaining the fairness and helps to eliminate starvation of lower priority class services. Bandwidth utilization of the system and fairness index of the resources are also encountered to validate the QoS provided by our proposed scheduling scheme.

💡 Research Summary

**
The paper addresses a long‑standing challenge in IEEE 802.16 (WiMAX) broadband wireless access networks: how to guarantee the quality‑of‑service (QoS) requirements of heterogeneous traffic while preserving fairness among service classes. WiMAX defines five QoS classes—Unsolicited Grant Service (UGS), extended real‑time Polling Service (ertPS), real‑time Polling Service (rtPS), non‑real‑time Polling Service (nrtPS), and Best‑Effort (BE)—each with distinct constraints on minimum bandwidth, maximum delay, and jitter. Existing schedulers, such as simple priority queuing, weighted round‑robin (WRR), and delay‑aware schemes like Modified Largest Weighted Delay First (M‑LWD), either fail to adapt quickly to traffic bursts or allow high‑priority traffic to starve lower‑priority flows, resulting in poor fairness and inefficient spectrum use.

The authors propose a novel QoS scheduler that operates in three logical stages: (1) Traffic Characterization, (2) Dynamic Weight Adjustment, and (3) Packet Transmission Decision. In the first stage, the scheduler continuously monitors per‑class queue length, average arrival rate, and the remaining time before each packet’s deadline expires. These metrics are refreshed every scheduling interval, providing a real‑time view of traffic dynamics.

In the second stage, the scheduler computes a dynamic weight for each class. The base weight reflects the class’s static priority, but it is multiplied by two adaptive factors: (i) a deadline urgency factor, which grows as the remaining time to deadline shrinks, and (ii) an occupancy factor, proportional to the current queue length. Consequently, a class that is about to miss its delay bound or that has accumulated a large backlog receives a higher weight for the upcoming round.

The third stage uses the updated weights in a Weighted Round‑Robin (WRR) arbitration. Slots are allocated proportionally to the weights, guaranteeing that urgent or backlogged traffic gets service promptly. Importantly, the algorithm monitors Jain’s Fairness Index (JFI) across all classes. If the JFI falls below a pre‑defined threshold, any unused bandwidth in the current frame is redistributed preferentially to lower‑priority classes (nrtPS and BE) to restore fairness. This “fairness‑driven residual allocation” prevents starvation without sacrificing overall throughput.

The authors evaluate the scheduler through extensive system‑level simulations of an IEEE 802.16e OFDMA cell. Four representative traffic mixes are used: VoIP (UGS), real‑time video (rtPS/ertPS), bulk file transfer (nrtPS), and web browsing (BE). Performance metrics include average packet delay, delay‑violation rate (DVR), aggregate throughput, bandwidth utilization, and JFI. Compared with a conventional priority‑queue scheduler and the M‑LWD scheme, the proposed method achieves:

• Delay Reduction – average end‑to‑end delay is lowered by roughly 25 % for real‑time flows; DVR for rtPS/ertPS falls below 0.5 % across all load levels.
• Throughput Improvement – total system throughput rises by about 12 % due to more efficient use of idle slots.
• Higher Bandwidth Utilization – the scheduler consistently occupies 85 %–90 % of the available spectrum, whereas the baseline often leaves 10 %–15 % idle.
• Fairness Preservation – JFI remains in the 0.96–0.99 range, indicating near‑perfect fairness; lower‑priority nrtPS and BE traffic experience negligible starvation.
• Starvation Elimination – the residual‑allocation mechanism ensures that even under heavy real‑time load, best‑effort users receive a minimum guaranteed share of resources.

The paper concludes that the combination of real‑time weight adaptation and fairness‑aware residual bandwidth redistribution constitutes an effective solution for meeting the dual objectives of QoS guarantee and equitable resource sharing in WiMAX networks. The authors argue that the same principles can be extended to newer broadband radio access technologies such as LTE‑Advanced and 5G NR, where heterogeneous service requirements and dense spectrum reuse are even more pronounced. Future work is suggested in three directions: (i) inter‑cell coordination to mitigate inter‑cell interference while preserving the proposed fairness model, (ii) incorporation of energy‑efficiency metrics into the weight calculation, and (iii) prototype implementation on a software‑defined radio platform to validate the algorithm under real‑world channel conditions.