面向智能手机的多源融合定位精度和能耗权衡机制研究

Localization accuracy and energy consumption are two key and correlated performance metrics for localization algorithms, and especially for smartphones with limited energy capacity and multiple location sensing abilities (e.g. GPS, WLAN and IMU), implementing multi-source fusion localization methods that satisfy both precision requirements and energy constraints is an urgent problem to be solved. This project intends to focus on trading off between the localization accuracy and energy consumption of the fusion localization based on GPS/WLAN/IMU. First, after comprehensively analyzing the localization errors of the WLAN-based fingerprinting method, systematic localization optimization methods and energy-saving strategies will be proposed, and critical algorithms for outdoor localization will be presented as well. Second, by in-depth analyzing the estimation error and energy consumption of pedestrian dead reckoning(PDR) algorithms based on IMU as well as their dependency on IMU sampling frequencies, the energy efficiency of the PDR algorithms will be significantly improved by selecting appropriate sampling frequencies. Last but not least, through investigating the error mechanism and the energy consumption rule of the fusion localization, the constraint model among the localization error, device duty cycle, sampling frequency, energy consumption and other factors will be established, so as to trade off between the localization accuracy and energy consumption of the fusion localization by adaptively scheduling the working status and the sampling frequencies of the devices embedded in smartphones. This project not only promotes the theory of smartphone localization, but also provides critical technical supports for energy-efficient localization services.

定位算法的精度与能耗是相互制约的两个关键性能指标，特别是对于具备GPS、WLAN和IMU等多种定位手段、能量受限的智能手机来说，实现满足不同精度需求和能耗约束的多源融合定位方法显得尤为重要，是一个亟待解决的难题。本项目拟围绕基于GPS/WLAN/IMU的多源融合定位精度和能耗权衡机制开展研究。首先，从理论上全面分析WLAN位置指纹定位的误差机理，形成系统的定位优化方法和能耗节约策略，提出适用于室外环境的关键算法。其次，深入剖析基于IMU的行人航迹推算算法的误差、能耗及其与IMU采样频率的依赖关系，通过合理选择采样频率提高算法的能效。最后，探索多源融合定位方法的误差影响机制和能耗变化规律，构建定位误差、设备占空比、采样频率、能耗等要素的约束关系模型，提出自适应的设备调度算法，解决融合定位的精度和能耗权衡问题。本项目不仅有助于完善智能手机的定位理论，而且为实现高能效的定位服务提供关键技术保障。

基于位置的服务（Location Based Services，LBS）是物联网和移动计算领域的热门研究方向之一。随着近二十年来技术的不断发展，例如GPS、WLAN、IMU等，定位手段越来越多样化。然而，每种定位技术都存在自身的局限性，很难单独在复杂多变的场景中提供普适的位置服务以满足各种各样的位置需求。为此，在大多数场景中，更多的是依靠不同定位技术的有效融合来提供精确、可靠的位置服务，但也不可避免地增加了定位能耗。因此，对于能量受限的智能手机来说，如何实现满足不同精度需求与能耗约束的多源融合定位成为一个亟待解决的难题。基于此背景，本项目以多源数据融合定位为出发点，探索了GPS/WLAN/IMU融合的定位方法，重点对WLAN指纹定位误差理论、基于IMU的PDR采样频率、精度、能耗的制约关系、以及三种技术相融合的误差与能耗权衡机制进行了全面而深入的研究。具体来说，一方面，本项目建立了WLAN指纹定位误差模型，深入分析了不同因素对指纹定位精度的影响，并从环境动态性、设备异质性、信道多样性、指纹库冗余性等方面对WLAN指纹定位性能进行了全面优化；另一方面，本项目深入研究了基于IMU的PDR中传感器频率对计步、航向估计精度和能耗的影响，设计了在没有精度损失情况下的低能耗计步、航向估计方法；在以上研究基础之上，本项目进一步探究了GPS/WLAN/IMU融合定位方法，提出了基于模糊逻辑的自适应的设备调度算法，实了现定位精度和能耗的有效权衡。本项目成果对于完善智能手机的定位理论和实现高能效的定位系统设计具有重要意义。

内容获取失败，请点击重试