Medium Access Control Protocols for Wireless Sensor Networks with Energy Harvesting

Fabio Iannello, Osvaldo Simeone, Umberto SpagnoliniIEEE Trans. Communications 2012 · DOI

TL;DR

This paper tackles the challenge of designing efficient Medium Access Control (MAC) protocols for wireless sensor networks (WSNs) powered by energy harvesting (EH). It introduces new performance metrics, "delivery probability" and "time efficiency," to balance data delivery with energy constraints, and analytically evaluates TDMA, Framed-ALOHA (FA), and Dynamic-FA (DFA) protocols under these new criteria.

What Problem Does This Paper Address?

Traditional wireless sensor networks (WSNs) are often designed with battery-powered (BP) devices, where network lifetime is the primary performance concern. However, the advent of energy harvesting (EH) technologies allows sensors to replenish their energy from the environment, potentially enabling perpetual operation. This shift introduces new design challenges: EH devices face unpredictable energy availability, which can lead to temporary energy shortages and impact short-term activities. This paper addresses the need for MAC protocols that can effectively manage this uncertainty, moving beyond traditional lifetime-centric designs to consider the trade-offs between delivering data reliably and efficiently utilizing harvested energy.

Methods

The authors propose a system-level analysis of MAC protocols for single-hop WSNs with an EH-capable sensor and a fusion center (FC). They introduce two key performance metrics:

  • Delivery Probability: Measures the likelihood of a sensor successfully delivering its data to the FC, considering both energy availability and successful transmission.
  • Time Efficiency: Quantifies the rate at which the FC collects data, essentially measuring how effectively the allocated time slots are utilized.

They then analyze three standard MAC protocols:

  • TDMA (Time Division Multiple Access): Each sensor has a pre-assigned slot, offering collision-free communication but potentially leading to wasted slots if sensors lack data or energy.
  • Framed-ALOHA (FA): Sensors contend for slots within frames. This is a simpler version of DFA without retransmissions.
  • Dynamic-FA (DFA): An extension of FA that allows for retransmissions within an IR, dynamically adjusting frame sizes based on estimated backlog (number of sensors with data to transmit).

The analysis relies on analytical frameworks, including Markov models, and makes simplifying assumptions like known backlog and large backlog approximations to derive performance metrics. A practical, low-complexity backlog estimation algorithm is also proposed.

Datasets

Not discussed in the paper. The evaluation is based on analytical derivations and simulations with system parameters.

Key Insights and Ideas

  • New Metrics for EH-WSNs: The traditional focus on network lifetime is insufficient for EH-WSNs. The proposed "delivery probability" and "time efficiency" metrics provide a more holistic view of performance by balancing data delivery success with the rate of data collection, crucial for energy-constrained systems.
  • TDMA's Trade-off: TDMA offers the highest delivery probability due to its collision-free nature but can be inefficient in terms of time efficiency, especially when energy harvesting rates are low or sensor activity is sparse.
  • FA and DFA's Adaptability: FA and DFA, being contention-based, offer more flexibility by dynamically adjusting frame sizes (in DFA) based on backlog estimates. This allows them to better adapt to varying energy availability and sensor activity, leading to improved time efficiency.
  • DFA's Retransmission Advantage: DFA's retransmission capability is particularly beneficial when harvesting rates are high, allowing it to recover from initial transmission failures and improve delivery probability compared to FA. However, at low harvesting rates, the advantage diminishes as many sensors are in energy shortage.
  • Backlog Estimation is Crucial: The performance of FA and DFA heavily relies on accurate backlog estimation. The proposed algorithm offers a practical solution for estimating backlog, which is essential for dynamic frame sizing.
  • Capture Effect Matters: The "capture effect" (where one successful transmission can occur even with multiple simultaneous transmissions) significantly impacts the performance of ALOHA-based protocols, making them more robust to collisions than a simple collision model would suggest.

Key Results

  • TDMA: Consistently achieves the highest asymptotic delivery probability across various harvesting rates. Its time efficiency is independent of the backlog estimation parameter 'p'.
  • FA and DFA: Show a trade-off between delivery probability and time efficiency that can be managed by adjusting the design parameter 'p' (related to frame length).
  • DFA vs. FA: DFA generally outperforms FA, especially at higher harvesting rates, due to its retransmission capabilities. At low harvesting rates, their performance is similar.
  • Backlog Estimation Validation: The analytical results are validated by simulations that compare performance with both known and estimated backlogs, demonstrating the efficacy of the proposed estimation algorithm.
  • SIR Threshold Impact: Lower SIR thresholds (indicating a higher probability of successful capture) improve the performance of ALOHA-based protocols (FA and DFA), while TDMA remains unaffected.

Limitations and Future Work

  • Simplifying Assumptions: The analytical framework relies on assumptions like known backlog and large backlog approximations. While validated by simulations, these assumptions might limit the direct applicability to all real-world scenarios.
  • Single-Hop WSNs: The analysis is focused on single-hop WSNs. Future work could extend this to multi-hop scenarios.
  • Energy Harvesting Model: The energy harvesting model is simplified. More complex and realistic EH models could be explored.
  • Routing for EH-WSNs: The paper focuses on MAC protocols. Future research could investigate routing protocols specifically designed for EH-WSNs.
  • Complex EH Dynamics: The paper assumes EH dynamics are slower than the IR duration. Investigating scenarios with faster EH dynamics could be a future direction.

Summary updated

BibTeX
@article{iannello2012medium,
  author = {Fabio Iannello and Osvaldo Simeone and Umberto Spagnolini},
  title = {{Medium Access Control Protocols for Wireless Sensor Networks with Energy Harvesting}},
  journal = {IEEE Trans. Communications},
  year = {2012},
  doi = {10.1109/tcomm.2012.030712.110089},
}
Figure