Adaptive DRX Scheme to Improve Energy Efficiency in LTE Networks with Bounded Delay

Abstract: The Discontinuous Reception (DRX) mechanism is commonly employed in current LTE networks to improve energy efficiency of user equipment (UE). DRX allows UEs to monitor the physical downlink control channel (PDCCH) discontinuously when there is no downlink traffic for them, thus reducing their energy consumption. However, DRX power savings are achieved at the expense of some increase in packet delay since downlink traffic transmission must be deferred until the UEs resume listening to the PDCCH. In this paper, we present a promising mechanism that reduces energy consumption of UEs using DRX while simultaneously maintaining average packet delay around a desired target. Furthermore, our proposal is able to achieve significant power savings without either increasing signaling overhead or requiring any changes to deployed wireless protocols.

• The paper introduces a coalesced DRX scheme that buffers packets at the eNB to extend UE sleep periods and improve energy efficiency without requiring protocol changes.
• It develops an analytical delay model and validates the energy-delay trade-off with simulations under varying traffic conditions including Poisson, dynamic, and real video streams.
• An adaptive algorithm is presented that tunes the queue threshold in real time to maintain target delay values while maximizing energy savings.

Adaptive DRX Scheme for Energy Efficiency in LTE Networks with Bounded Delay

Introduction and Motivation

The paper addresses the challenge of improving energy efficiency in LTE/LTE-A user equipment (UE) through Discontinuous Reception (DRX) mechanisms, while maintaining bounded packet delay. DRX allows UEs to periodically enter low-power states by discontinuously monitoring the Physical Downlink Control Channel (PDCCH), reducing energy consumption at the cost of increased downlink packet delay. The trade-off between energy savings and delay is governed by DRX parameters, which are typically statically configured per UE. However, static configurations are suboptimal under varying traffic conditions, and dynamic reconfiguration via RRC signaling introduces significant overhead, especially in dense deployments.

Coalesced DRX: Mechanism and Rationale

The proposed scheme, termed "coalesced DRX," leverages packet coalescing at the eNB to further extend UE sleep periods without modifying DRX parameters or increasing signaling. Instead of transmitting downlink packets as soon as the UE becomes active, the eNB buffers packets until a configurable queue threshold $Q_w$ is reached, or a maximum waiting time $W_{\max}$ is exceeded. This approach increases the fraction of time UEs spend in low-power mode, directly improving energy efficiency.

The key insight is that the eNB can control the trade-off between delay and energy savings by adjusting $Q_w$, without requiring protocol changes or additional signaling. The mechanism is compatible with existing LTE/LTE-A deployments and only requires minor modifications at the eNB.

Analytical Delay Model

A detailed queueing-theoretic model is developed to quantify the impact of coalesced DRX on average packet delay. The model considers general arrival and service time distributions, and explicitly accounts for the DRX cycle structure, inactivity timers, and the coalescing threshold. The analysis introduces the $\gamma$ factor, representing the inverse of the fraction of idle time at the eNB with DRX enabled, and derives closed-form expressions for the average queueing delay $\mathrm{E}[W]$ as a function of $Q_w$, traffic parameters, and DRX configuration.

For Poisson arrivals, the model simplifies, yielding tractable expressions for $\mathrm{E}[W]$ and enabling efficient computation of the delay-energy trade-off. The model is validated via simulation, demonstrating high accuracy across a range of traffic loads and DRX settings.

Adaptive Queue Threshold Algorithm

Recognizing that a fixed $Q_w$ is suboptimal under dynamic traffic, the paper proposes an adaptive algorithm to tune $Q_w$ in real time. The algorithm operates as a closed-loop controller: after each coalescing cycle, the eNB measures the average queueing delay $\widehat{W}[i]$ and updates $Q_w$ according to the deviation from a target delay $W^*$. The update rule is:

$$Q_w[i+1] = Q_w[i] + 2\widehat{\lambda}(W^* - \widehat{W}[i])$$

where $\widehat{\lambda}$ is the estimated arrival rate. The algorithm enforces bounds on $Q_w$ to ensure that the maximum delay $W_{\max}$ is not exceeded. Stability analysis in the appendix shows that the controller is stable for practical parameter ranges.

This adaptive approach allows the system to maintain average delay near the target $W^*$ while maximizing energy savings, without incurring signaling overhead or requiring protocol changes.

Simulation Results

Extensive simulations are conducted to evaluate the performance of coalesced DRX and the adaptive algorithm under various traffic models:

• Poisson Traffic: Coalesced DRX with adaptive $Q_w$ achieves significant energy savings compared to standard DRX, while maintaining average delay close to the target. At low traffic loads, small $Q_w$ values are selected to avoid excessive delay, while at high loads, larger $Q_w$ values are used to maximize energy savings.
• Dynamic Traffic: The adaptive algorithm rapidly tracks changes in traffic load, adjusting $Q_w$ to maintain delay targets and optimize energy savings.
• Self-Similar (Pareto) Traffic: The scheme remains effective under heavy-tailed traffic, with the adaptive controller maintaining delay bounds and achieving higher energy savings than standard DRX.
• Real Video Streaming Traces: Using YouTube and SopCast traces, the scheme demonstrates robust performance, achieving notable energy savings and bounded delay across diverse real-world traffic patterns.

Implementation Considerations

The proposed scheme is designed for practical deployment:

• eNB Modifications: Only minor changes are required at the eNB to implement queue thresholding and the adaptive controller. No changes to UE or wireless protocols are necessary.
• Parameter Configuration: The scheme requires configuration of two parameters per UE: target average delay $W^*$ and maximum delay $W_{\max}$. These can be set based on application requirements or UE power preference indications (PPI) as per 3GPP standards.
• Signaling Overhead: Unlike prior adaptive DRX schemes, the proposed method does not require RRC reconfiguration, avoiding additional signaling and associated energy costs.
• Computational Overhead: The adaptive algorithm is lightweight, requiring only per-UE queueing delay and arrival rate estimation, which can be efficiently implemented.

Comparison with Related Work

Most prior DRX adaptation schemes rely on dynamic reconfiguration of DRX parameters, incurring RRC signaling overhead and increased processing at both eNB and UE. The only comparable approach, CDA-DRX, requires protocol modifications and negotiation of new parameters, limiting deployability. In contrast, the coalesced DRX scheme achieves similar or better energy-delay trade-offs without protocol changes or signaling overhead, and with simpler configuration.

Implications and Future Directions

The coalesced DRX scheme provides a practical, low-overhead method for improving UE energy efficiency in LTE/LTE-A networks while maintaining bounded delay. Its compatibility with existing deployments and minimal configuration requirements make it attractive for immediate adoption. The analytical model enables precise tuning of delay-energy trade-offs, and the adaptive algorithm ensures robust performance under dynamic and heterogeneous traffic.

Potential future developments include:

• Integration with DRX-aware Scheduling: Combining coalesced DRX with DRX-aware schedulers at the eNB could further optimize resource allocation under high load.
• Extension to 5G NR: The principles of coalesced DRX are applicable to 5G NR, where similar power-saving mechanisms exist.
• Multi-flow and QoS Differentiation: Extending the scheme to support per-flow delay targets and differentiated QoS could enhance support for mixed traffic scenarios.
• Machine Learning-based Adaptation: Incorporating predictive models for traffic patterns could further improve adaptation speed and accuracy.

Conclusion

The paper presents a coalesced DRX scheme and an adaptive queue threshold algorithm that together enable significant improvements in UE energy efficiency in LTE/LTE-A networks, with bounded and configurable packet delay. The approach avoids signaling overhead, requires minimal changes at the eNB, and is validated through rigorous analysis and simulation under diverse traffic conditions. The scheme is well-suited for practical deployment and provides a foundation for further enhancements in energy-efficient wireless access.