[Paper Review] Energy Efficient Medium Access with Interference Mitigation in LTE Femtocell Networks

Ying Wang, Xiangming Dai+2 more|arXiv (Cornell University)|Aug 6, 2015

Advanced MIMO Systems Optimization3 citations

TL;DR

This paper proposes two distributed, interference-aware algorithms—Greedy Interference-based (IG) and Fast Interference-based (FIG)—for energy-efficient medium access in LTE femtocell networks by jointly optimizing user equipment (UE) association and OFDMA scheduling. The algorithms achieve up to 150% higher utility and over 2x better energy efficiency than benchmark methods, while maintaining low X-2 interface signaling and meeting QoS requirements for both GBR and Non-GBR traffic.

ABSTRACT

With the rapidly increasing number of deployed LTE femtocell base stations (FBS), energy consumption of femtocell networks has become a serious environmental issue. Therefore, energy-efficient protocols are needed to balance the trade-off between energy saving and bandwidth utilization. The key component of the desired protocol to prevent both energy and bandwidth waste is interference mitigation, which, nevertheless, most previous work has failed to properly consider. To this end, in this paper, we manipulate user equipment (UE) association and OFDMA scheduling with a combination of interference mitigation. Recognizing the NP-hardness of the problem, we propose two distributed algorithms with guaranteed convergence. Extensive simulations show that our algorithms outperform the alternative algorithms in multiple metrics such as utility, power consumption, and convergence speed.

Motivation & Objective

• Address the growing energy consumption of large-scale LTE femtocell networks, which poses environmental and operational challenges.
• Overcome the limitations of prior work that neglect interference, QoS differentiation, or X-2 interface latency in energy-efficient designs.
• Develop a practical, distributed MAC protocol that minimizes coordination overhead while balancing energy efficiency and bandwidth utilization.
• Jointly optimize UE association and OFDMA scheduling to manage interference and meet QoS requirements for both GBR and Non-GBR traffic.
• Introduce a tunable power weight parameter to balance utility maximization and power minimization in heterogeneous traffic environments.

Proposed method

• Formulate the problem as a joint optimization of UE association and OFDMA scheduling to minimize energy consumption and maximize utility under QoS constraints.
• Propose two greedy, distributed algorithms—IG and FIG—that converge reliably and reduce X-2 interface signaling by using probabilistic OFDMA tile allocation.
• Incorporate interference mitigation by modeling multi-cell multi-link interference in the OFDMA resource allocation process.
• Use a weighted utility function combining spectral efficiency and energy consumption, with a tunable power weight parameter to balance trade-offs.
• Implement a practical MAC protocol that enables distributed decision-making with minimal coordination, avoiding centralized or high-latency coordination.
• Validate the approach via extensive simulations comparing IG, FIG, and benchmark algorithms (LA, SA) across utility, power consumption, QoS, and energy efficiency metrics.

Experimental results

Research questions

• RQ1 How can energy efficiency and bandwidth efficiency be jointly optimized in LTE femtocell networks while managing inter-cell interference?
• RQ2 What is the impact of interference mitigation on utility and energy efficiency when QoS requirements for both GBR and Non-GBR traffic are enforced?
• RQ3 How does the proposed distributed MAC protocol reduce X-2 interface signaling compared to centralized or coordination-heavy alternatives?
• RQ4 To what extent can distributed algorithms approximate the performance of centralized optimal solutions in terms of utility and energy efficiency?
• RQ5 How does tuning the power weight parameter affect the trade-off between utility and energy consumption in heterogeneous traffic scenarios?

Key findings

• IG and FIG improved utility by 150% over the benchmark LA algorithm due to effective interference mitigation and efficient resource allocation.
• IG and FIG achieved energy efficiency over 200% higher than LA, approaching the performance of the centralized upper-bound SA algorithm.
• The GBR reject ratio for IG and FIG was less than 9%, which is only 1/6th of LA’s, demonstrating strong QoS support for real-time services.
• Non-GBR utility for IG and FIG was nearly three times higher than LA, with only a 14% drop compared to SA, indicating strong performance for elastic traffic.
• When the power weight ω was set to 2.0, IG and FIG achieved 21.30 ± 0.96 and 21.27 ± 0.95 energy efficiency units, respectively, surpassing SA’s 16.27 ± 0.83, showing superior performance at high energy priority.
• IG and FIG showed higher standard deviation than SA due to greedy search variability, but still delivered consistent and superior performance across multiple metrics.