Analysis of Latency Performance of Bluetooth Low Energy (BLE) Networks

Keuchul Cho, Woojin Park, Moonki Hong, Gisu Park, Wooseong Cho, Jihoon Seo, Kijun Han2015 · DOI

TL;DR

This paper tackles the challenge of accurately modeling Bluetooth Low Energy (BLE) device discovery, a crucial but complex process due to its unique advertising channels and parameter flexibility. The authors develop an analytical model and validate it with simulations, providing insights into how various parameters impact discovery probability and latency, which is essential for optimizing BLE device performance.

What Problem Does This Paper Address?

Bluetooth Low Energy (BLE) is a popular short-range wireless technology known for its low power consumption, making it ideal for IoT devices and wearables. A fundamental aspect of BLE is its "device discovery" process, where devices advertise their presence and scanners search for them. While classical Bluetooth discovery has been extensively studied, BLE's discovery mechanism is fundamentally different, utilizing three dedicated advertising channels and offering a wide range of configurable parameters. This flexibility, while beneficial for customization, makes accurate performance analysis challenging. Existing studies on BLE discovery are often not thorough enough, leaving a gap in understanding how parameter settings influence critical performance metrics like discovery probability and latency. This paper aims to fill this gap by developing a new, accurate analytical model for BLE device discovery.

Methods

The core of this paper is the development of an analytical model for BLE device discovery. The authors first define six key events that describe the conditions for successful discovery between an advertiser and a scanner. They then build upon these events to derive probabilities of successful discovery for different network scenarios, starting with a simple 1:1 network and progressing to more complex M:N networks.

A key aspect of their approach is considering both continuous and discontinuous scanning modes. They introduce a "duty cycle" (p = tsw/tsi) to quantify how long a BLE device spends scanning within a given interval. The model accounts for the timing of advertising and scanning events, including the transmission times of various BLE packets (ADV_IND, SCAN_REQ, SCAN_RSP) and inter-frame spaces.

To validate their analytical model, the authors developed a BLE simulation program that adheres to the BLE specification. They then compared the results from their analytical model with those obtained from extensive simulations across various parameter settings.

Datasets

Not discussed in the paper. The paper focuses on developing and validating an analytical model through simulation, rather than using pre-existing datasets.

Key Insights and Ideas

  • BLE Discovery is Inherently Probabilistic and Collision-Prone: Unlike some other wireless technologies, BLE discovery relies on probabilistic rendezvous on one of three advertising channels. This, combined with the use of tiny frames and the lack of channel state examination before transmission, leads to a significant probability of collisions, especially as the number of devices increases.
  • Parameter Tuning is Crucial for Performance: The wide range of configurable parameters in BLE discovery (e.g., AdvInterval, ScanWindow) offers significant flexibility but also means that inappropriate settings can severely degrade performance, leading to low discovery probabilities and high latencies.
  • Discovery Latency Grows Exponentially with Device Count: As the number of BLE devices in a network increases, the mean discovery latency can grow exponentially. This is a direct consequence of increased contention and collisions on the advertising channels.
  • ScanWindow (tsw) is a Key Driver of Latency: The duration of the scanning window (tsw) has a more significant impact on discovery latency than the advertising interval per channel (twa). Increasing tsw generally leads to lower latency, up to a certain point, as it allows more time for scanning and potential rendezvous.
  • AdvInterval (tAI) Directly Impacts Latency: The discovery latency is shown to be directly proportional to the AdvInterval (tAI). This is because advertisers initiate their advertising intervals based on tAI, making it a primary factor in how quickly a discovery event can be initiated.

Key Results

  • The developed analytical model for BLE device discovery probability and latency was shown to closely match simulation results across a wide range of parameter settings.
  • The study quantifies the low success probability of BLE discovery, even in relatively small networks (e.g., ~0.3 in a 1:1 network), attributing this to the three-channel synchronization and potential collisions.
  • Increasing the number of scanners (N) or advertisers (M) significantly decreases discovery probability and increases discovery latency, demonstrating the impact of contention.
  • The ScanWindow (tsw) was found to have a noticeable impact on discovery probability, with an increase in tsw generally leading to higher probability up to a point.
  • The AdvInterval (tAI) was found to have a significant impact on discovery latency, with higher tAI values leading to higher latency.
  • The sensitivity analysis revealed that discovery latency is highly sensitive to tAI, moderately sensitive to tsw, and less sensitive to twa.

Limitations and Future Work

  • Focus on 1:1 and M:N Networks: While the paper covers M:N networks, the analysis primarily focuses on the interaction between a single advertiser and one or more scanners, or vice-versa. More complex scenarios with multiple advertisers and scanners interacting simultaneously might require further investigation.
  • Idealized Assumptions: The model likely makes some idealized assumptions, such as perfect synchronization within scanning windows and no external interference, which might not hold true in real-world, highly congested environments.
  • Future Work: The authors suggest that their work is a foundational step, and further research could explore more complex network topologies, different BLE advertising modes (directed vs. undirected), and the impact of external interference on discovery performance. The flexibility of BLE parameters also suggests future work in adaptive discovery mechanisms that dynamically adjust parameters based on network conditions.

Summary updated

BibTeX
@misc{cho2015analysis,
  author = {Keuchul Cho and Woojin Park and Moonki Hong and Gisu Park and Wooseong Cho and Jihoon Seo and Kijun Han},
  title = {{Analysis of Latency Performance of Bluetooth Low Energy (BLE) Networks}},
  year = {2015},
  doi = {10.3390/s150100059},
}
Figure