Frequency and Time Domain Packet Scheduling Based on Channel Prediction with Imperfect CQI in LTE

Yongxin Wang, Kumbesan Sandrasegaran, Xinning Zhu, Jingjing Fei, Xiaoying Kong, Cheng Lin

Abstract

Channel-dependent scheduling of transmission of data packets in a wireless system is based on measurement and feedback of the channel quality. To alleviate the performance degradation due to simultaneous multiple imperfect channel quality information (CQI), a simple and efficient packet scheduling (PS) algorithm is developed in downlink LTE system for real time traffic. A frequency domain channel predictor based on Kalman filter is first developed to restore the true CQI from erroneous channel quality feedback. Then, a time domain grouping technique employing the joint of Proportional Fair (PF) and Modified Largest Weighted Delay First (M-LWDF) algorithms is used. It was proved this proposal achieves better performance in terms of system throughput and packet loss ratio by simulation results.

Introduction

In Long Term Evolution (LTE) systems, the evolved NodeB (eNB) must make rapid resource allocation decisions every millisecond to maintain high Quality of Service (QoS). These decisions rely heavily on Channel Quality Information (CQI) reported by user equipment (UE). However, in practical deployments, this feedback is often "imperfect" due to transmission delays, environmental noise, and intermittent unavailability. When scheduling is based on such stale or erroneous data, system performance degrades significantly, particularly for real-time applications like video conferencing or gaming. This paper proposes a robust packet scheduling framework that compensates for these imperfections through predictive modeling and an alternating scheduling strategy.

Proposed Methodology

The proposed solution addresses channel imperfections through a two-stage approach involving frequency-domain prediction and time-domain grouping.

Frequency Domain Channel Prediction

To mitigate the impact of delayed or noisy CQI, the authors implement a Kalman Filter-based channel predictor. This filter operates through a recursive estimation-correction loop to track the Signal-to-Interference-plus-Noise Ratio (SINR). By modeling the channel state using a Taylor series expansion, the filter estimates not only the current value but also the rate of change and gradient of the channel quality. This allows the eNB to maintain an accurate estimate of the channel state even during periods of feedback "blackout."

Flowchart of the Kalman Filter process, illustrating the recursive Estimation and Correction steps used for channel prediction.

Time Domain Grouping Technique

In the time domain, the framework employs a grouping technique that alternates between two established scheduling algorithms across Transmission Time Intervals (TTIs). By dividing TTIs into odd and even groups, the system utilizes:

  • Proportional Fair (PF): To balance overall system throughput with user fairness.
  • Modified Largest Weighted Delay First (M-LWDF): To prioritize users with packets approaching their delay deadlines.

This alternating approach provides a balanced performance profile—maintaining high throughput while managing packet delay—without the high computational complexity associated with joint optimization algorithms.

Simulation Environment

The effectiveness of the proposed algorithm was evaluated using a system-level simulation modeled after 3GPP LTE standards. The simulation considers a high-mobility scenario with users moving at 120 km/h, creating a challenging environment for channel prediction due to rapid fading.

Table of system parameters defining the LTE simulation environment, including cellular layout, bandwidth, and mobility settings.

Key impairments introduced in the simulation include a 3 ms CQI feedback delay and periodic CQI unavailability every 10 ms to simulate realistic network conditions.

Performance Evaluation

The proposed algorithm was compared against standard Frequency Domain Proportional Fair (FD-PF) and FD-M-LWDF implementations. Results demonstrate significant improvements under imperfect CQI conditions:

  • Packet Loss Ratio (PLR): In a high-load scenario with 70 users, the proposed method achieved a 116.94% improvement over the PF algorithm and a 6.31% improvement over M-LWDF.
  • System Throughput: The proposed method reached a throughput of 6.1792 Mbps, outperforming both PF (5.6921 Mbps) and M-LWDF (5.9174 Mbps).
  • System Capacity: The framework supports a higher number of users before reaching the QoS failure threshold (typically a PLR of $10^{-6}$ for real-time traffic).

Conclusion

The integration of a Kalman Filter for channel prediction and a time-domain alternating scheduling strategy offers a computationally efficient and robust solution for LTE packet scheduling. The Kalman filter's ability to maintain a state model during CQI blackouts is particularly valuable in high-mobility environments. Future research may explore the performance of this framework under mixed traffic types and more complex inter-cell interference scenarios.

Summary updated

Abstract

Channel-dependent scheduling of transmission of data packets in a wireless system is based on measurement and feedback of the channel quality. To alleviate the performance degradation due to simultaneous multiple imperfect channel quality information (CQI), a simple and efficient packet scheduling (PS) algorithm is developed in downlink LTE system for real time traffic. A frequency domain channel predictor based on Kalman filter is first developed to restore the true CQI from erroneous channel quality feedback. Then, a time domain grouping technique employing the joint of Proportional Fair (PF) and Modified Largest Weighted Delay First (M-LWDF) algorithms is used. It was proved this proposal achieves better performance in terms of system throughput and packet loss ratio by simulation results.
BibTeX
@article{wang2013frequency,
  author = {Yongxin Wang and Kumbesan Sandrasegaran and Xinning Zhu and Jingjing Fei and Xiaoying Kong and Cheng Lin},
  title = {{Frequency and Time Domain Packet Scheduling Based on Channel Prediction with Imperfect CQI in LTE}},
  journal = {International Journal of Wireless \& Mobile Networks},
  year = {2013},
  doi = {10.5121/ijwmn.2013.5412},
}
Figure