Modeling Battery Behavior on Sensory Operations for Context-Aware Smartphone Sensing

Ozgur Yurur, Chi Harold Liu, Wilfrido Alejandro MorenoSensors 2015 · OpenReview

TL;DR

This paper tackles the critical issue of battery drain in context-aware smartphone sensing by developing a novel model that accounts for battery non-linearities and sensor operation patterns. By integrating a kinetic battery model (KiBaM) with a Markov reward process, they propose methods to optimize sensor usage for extended battery life without sacrificing application accuracy.

What Problem Does This Paper Address?

Modern smartphones are packed with sensors, enabling sophisticated context-aware applications that can proactively assist users. However, the continuous operation of these sensors, especially in demanding applications like human activity recognition, leads to rapid battery depletion. This is a significant bottleneck for mobile computing, limiting the usability and longevity of these devices. Existing battery models often fail to capture the complex, non-linear behavior of batteries under varying loads, and they don't adequately account for how different sensor operation strategies (like sampling frequency and duty cycle) impact battery discharge. This paper aims to bridge this gap by providing a more accurate model that links sensor activity to battery behavior, enabling better energy management.

Methods

The authors employ a multi-faceted approach:

  • Kinetic Battery Model (KiBaM): They adopt and extend the KiBaM, a well-established analytical model that represents battery charge as being distributed between "available" and "bound" wells. This model is crucial for capturing non-linear battery behaviors like the rate capacity effect and recovery effect, which are often overlooked by simpler models.
  • Sensor Operation Modeling: They analytically model the energy consumption of smartphone sensors, using the accelerometer as a case study. This involves defining parameters like duty cycle (DC) and sampling frequency (fs) to characterize sensor activity.
  • Interrupt Poisson Process (IPP): For modeling sensor operations, they utilize the IPP, a special case of Markov-modulated Poisson processes, to represent the ON-OFF nature of sensor sampling.
  • Markov Reward Process: A Markov reward process is integrated to create energy consumption profiles for different sensor operation settings. Each profile is associated with an "accumulated reward" representing the total energy cost.
  • Three State Transition Methods: They propose three distinct methods to dynamically adjust sensor operation parameters (DC and fs) based on application accuracy feedback, aiming to balance power consumption and recognition performance. These methods are applied within a human activity recognition (HAR) context.

Datasets

Not explicitly discussed in the paper. The authors mention using a Blackberry RIM Storm II 9550 smartphone for experiments and refer to technical details of the ADXL346 accelerometer.

Key Insights and Ideas

  • Battery Non-Linearities Matter: The paper strongly emphasizes that battery behavior is not linear. The "recovery effect," where a battery can regain some capacity after periods of low load, is a key insight that can be leveraged for energy savings.
  • Sensor Settings Directly Impact Battery Life: Different combinations of sampling frequency and duty cycle for sensors can lead to vastly different battery depletion rates. Aggressive sampling (high frequency, high duty cycle) drains the battery much faster.
  • Dynamic Adjustment is Key: Simply setting fixed sensor parameters is suboptimal. Dynamically adjusting these parameters based on real-time application needs and observed battery behavior is crucial for achieving significant energy savings.
  • Trade-off Between Accuracy and Power: There's an inherent trade-off between sensor sampling aggressiveness (leading to higher accuracy) and power consumption. The proposed methods aim to find an optimal balance.

Key Results

  • The proposed model, incorporating KiBaM and Markov reward processes, effectively links sensor operation patterns to battery discharge.
  • Experiments with the accelerometer sensor demonstrated that aggressive sampling methodologies (e.g., 100% duty cycle, 100 Hz sampling) cause the battery to deplete significantly faster.
  • Lower duty cycles were found to enhance the battery recovery effect, prolonging battery life.
  • The three proposed state transition methods for dynamic sensor adjustment showed varying degrees of effectiveness. Method 3, which allows simultaneous adjustment of DC and fs, achieved the highest efficiency, offering up to 50% enhancement in power consumption with only a 10% decrease in recognition accuracy.

Limitations and Future Work

  • Generalizability of Sensor Models: While the accelerometer is modeled extensively, the paper suggests that other sensors with similar operation methods (microphone, gyroscope, etc.) can be modeled similarly, but this requires further validation.
  • Parameter Verification: The paper notes that parameters like rs and ri (stable energy consumption rates) are difficult to verify through smartphone programming alone, relying on analytical models.
  • Future Work: The authors suggest that their work can lead to discovering optimal energy reduction strategies and contribute to the continual improvement of context-aware mobile services. Further research could explore more complex sensor interactions and a wider range of applications.

Summary updated

Abstract

Energy consumption is a major concern in context-aware smartphone sensing. This paper first studies mobile device-based battery modeling, which adopts the kinetic battery model (KiBaM), under the scope of battery non-linearities with respect to variant loads. Second, this paper models the energy consumption behavior of accelerometers analytically and then provides extensive simulation results and a smartphone application to examine the proposed sensor model. Third, a Markov reward process is integrated to create energy consumption profiles, linking with sensory operations and their effects on battery non-linearity. Energy consumption profiles consist of different pairs of duty cycles and sampling frequencies during sensory operations. Furthermore, the total energy cost by each profile is represented by an accumulated reward in this process. Finally, three different methods are proposed on the evolution of the reward process, to present the linkage between different usage patterns on the accelerometer sensor through a smartphone application and the battery behavior. By doing this, this paper aims at achieving a fine efficiency in power consumption caused by sensory operations, while maintaining the accuracy of smartphone applications based on sensor usages. More importantly, this study intends that modeling the battery non-linearities together with investigating the effects of different usage patterns in sensory operations in terms of the power consumption and the battery discharge may lead to discovering optimal energy reduction strategies to extend the battery lifetime and help a continual improvement in context-aware mobile services.
BibTeX
@inproceedings{yurur2015modeling,
  author = {Ozgur Yurur and Chi Harold Liu and Wilfrido Alejandro Moreno},
  title = {{Modeling Battery Behavior on Sensory Operations for Context-Aware Smartphone Sensing}},
  booktitle = {Sensors},
  year = {2015},
  url = {https://openreview.net/forum?id=wEj6F5-05U},
}
Figure