Introduction
As 5G networks evolve, they increasingly rely on Heterogeneous Networks (HetNets)—a mix of large "umbrella" macro-cells and smaller, low-power access points (APs) such as WiFi or femtocells. Simultaneously, there is a push for "green" communication, where mobile users (MUs) are powered by Energy Harvesting (EH) from renewable sources like solar or wind. This paper addresses the optimization of uplink transmissions for these energy-constrained users in a multi-channel environment.
System Architecture
The authors consider a scenario where an EH-powered mobile user has a dedicated private channel to a base station (BS) but can also attempt to access a shared "common" channel at a nearby small-cell access point.
Accessing the common channel is stochastic; the user might fail to secure it due to congestion. This creates a fundamental decision problem: when is the best time to stop "saving" energy and start "transmitting" data? While waiting longer allows for more energy accumulation and a better chance of finding a high-quality channel, it simultaneously reduces the total time available for data transfer.
Methodology: The Save-Then-Transmit Protocol
The proposed opportunistic transmission scheme is based on a "save-then-transmit" protocol. In this model, the MU harvests energy over several time slots and then utilizes the entire accumulated battery to transmit in a single burst.
To optimize this timing, the authors formulate throughput maximization as a "rate-of-return" optimal stopping problem. Key methodological features include:
- Markovian Modeling: The framework accounts for time-correlated energy arrivals and channel states, providing a more realistic representation than independent and identically distributed (i.i.d.) models.
- Optimal Stopping Theory: The authors prove that the optimal strategy follows a threshold-based structure. In general Markovian settings, this threshold is time-dependent, while in simpler i.i.d. settings, it simplifies to a fixed value.
- Benchmarking: The EH-based results are compared against conventional power supplies (using "water-filling" power allocation) and "best-effort" delivery (transmitting immediately without saving).
Key Insights and EH Diversity
A central contribution of this work is the concept of "EH diversity." While randomness in energy arrival is typically viewed as a hurdle, the authors demonstrate that higher variance in energy arrival can actually improve throughput. If the energy supply is "bursty," an intelligent opportunistic scheme can wait for high-energy bursts to transmit at significantly higher rates, effectively "harvesting" the randomness itself.
Furthermore, the decision to transmit is governed by a simple rule: if the potential transmission rate (determined by current energy and channel quality) exceeds a calculated threshold, the user transmits; otherwise, they continue harvesting.
Results and Performance
Numerical simulations using synthetic data for Markovian channel gains and two-state EH models ("GOOD" vs. "BAD" arrival states) yielded several key findings:
- Superiority over Best-Effort: The opportunistic scheme consistently outperforms "best-effort" delivery, particularly when channel conditions are highly variable.
- Efficiency: The EH-based opportunistic scheme achieves approximately 70% of the throughput of a system with a conventional, unlimited power supply under the same average power constraint.
- HetNet Synergy: Accessing the common channel—even if only occasionally successful—strictly increases average throughput. The benefit is most pronounced when there is a significant disparity between private and common channel conditions.
- Saving Time: The authors proved that as the probability of securing the common channel increases, the average time a user spends "saving" energy decreases polynomially.
Limitations and Future Work
While the "save-then-transmit" approach is mathematically tractable, it is inherently suboptimal compared to continuous power control strategies. Additionally, the current model assumes negligible circuit power for sensing and processing, which may be significant for ultra-low-power EH devices. Future research could also explore multi-user coordination and game-theoretic interactions between multiple opportunistic users competing for the same common channel.
Summary updated