TL;DR
This paper tackles the challenge of energy-efficient backhauling for the dense small cell networks envisioned for 5G. They propose a novel scheduling and power control scheme that leverages concurrent transmissions in the millimeter-wave (mmWave) band to significantly reduce energy consumption while maintaining high throughput.
What Problem Does This Paper Address?
The advent of 5G mobile networks relies heavily on heterogeneous cellular networks (HCNs) with a dense deployment of small cells. While this architecture promises higher data rates, it creates a significant challenge for the backhaul network connecting these small cells. Wired solutions are often too costly and inflexible, making wireless backhaul, particularly in the millimeter-wave (mmWave) band, a necessity. mmWave offers high bandwidth but suffers from high propagation loss, necessitating directional beamforming. This directional nature, however, enables "spatial reuse" – allowing multiple links to transmit concurrently with minimal interference. The core problem this paper addresses is how to effectively schedule these concurrent mmWave backhaul transmissions and control their power to minimize overall energy consumption without sacrificing network performance.
Methods
The authors propose an energy-efficient mmWave backhauling scheme built upon two key components:
- Maximum Independent Set (MIS)-based Scheduling: They first model the interference relationships between potential backhaul links as a "contention graph." An edge exists between two links if they interfere significantly (above a defined threshold). The goal is to schedule as many non-interfering links as possible concurrently. Since finding the exact MIS is computationally expensive, they employ a greedy approximation algorithm based on minimum degree to find near-optimal sets of concurrent transmissions for each time slot (called Channel Time Allocations or CTAs).
- Power Control Algorithm: Once concurrent transmissions are scheduled for a given time slot, a power control algorithm is used. This algorithm determines the transmission power for each flow within that slot. The key idea is that by allocating more CTAs to a flow (due to concurrent transmissions), its transmission rate can be reduced, which in turn allows for a significant reduction in transmission power, leading to lower energy consumption. The algorithm aims to ensure that each flow still meets its required throughput.
The problem is formally framed as a mixed-integer nonlinear programming (MINLP) problem, aiming to minimize total energy consumption.
Datasets
Not discussed in the paper. The authors rely on extensive simulations with generated traffic patterns and system parameters.
Key Insights and Ideas
- Concurrent Transmission is Key to Energy Savings: The fundamental insight is that by enabling multiple mmWave backhaul links to transmit simultaneously (spatial reuse), we can allocate more time resources to each individual link. This allows for reduced transmission power, which is the primary driver of energy savings.
- The Interference Threshold is a Critical Parameter: The effectiveness of concurrent transmissions hinges on managing interference. The paper introduces an "interference threshold" to define when links are considered to interfere too much to be scheduled together. The choice of this threshold significantly impacts both energy consumption and throughput, and its optimal value is shown to be dependent on various system conditions.
- Balancing Throughput and Energy: The proposed scheme explicitly aims to maintain the throughput achieved by traditional serial TDMA while drastically reducing energy. This is achieved by intelligently trading transmission time for lower transmission power.
- MIS Approximation is Sufficient: While finding the true Maximum Independent Set is NP-hard, the paper demonstrates that a greedy approximation algorithm provides a good balance between performance and computational complexity for practical scheduling.
Key Results
- Significant Energy Reduction: The proposed scheme consistently achieves the lowest energy consumption compared to serial TDMA and a concurrent transmission scheme that uses maximum transmission power (CTFP). For instance, at traffic load 5, it saves about 31.8% and 65.5% energy compared to TDMA and CTFP, respectively.
- Higher Network Throughput: The scheme also demonstrates higher network throughput than serial TDMA, and in some scenarios, it approaches the throughput of CTFP while consuming significantly less energy.
- Superior Energy Efficiency: Energy efficiency (throughput per unit of energy) is significantly higher with the proposed scheme, often by orders of magnitude when plotted on a logarithmic scale.
- Threshold Optimization is Crucial: The paper shows that the optimal interference threshold is not fixed and varies with traffic load, BS distribution, and maximum transmission power. Selecting an appropriate threshold is vital for maximizing energy savings and throughput.
Limitations and Future Work
- Approximation Algorithm: While the MIS approximation is effective, it's not guaranteed to find the absolute optimal schedule.
- Interference Threshold Dependency: The optimal interference threshold is scenario-dependent, requiring careful tuning for different network conditions.
- Future Work: The authors suggest further investigation into the optimized choice of the interference threshold under various system parameters. They also implicitly point towards the need for adaptive threshold selection in dynamic network environments.
Summary updated