【作者】 蔡超；

【导师】 李青侠； 蒋洪波；

【作者基本信息】 华中科技大学 ， 信息与通信工程， 2019， 博士

【摘要】 基于声音信号的室内定位技术由于其在实际部署中的成本较低、定位精度高等因素受到了广泛的关注。研究基于声信号的低延迟室内定位技术不仅能让用户及时获取当前位置,提升用户体验,且能实现动态跟踪,为室内导航、路径规划等新应用提供关键技术支持。因此研究低延迟的室内定位技术有着较高的应用价值。然而,现有的基于声音信号的异步室内定位技术在动态信道下,即在通信距离、收发设备、运动状态上均有变化的信道中,存在定位延迟高的问题,这一问题背后的技术原因为:1.动态信道下声音信号的检测不可靠引入的延迟。不可靠的信号检测容易造成信标信号的误检、漏检或出现较大的时间误差而被滤除,因此,要收集足够多的信标信号用于定位时,需要较多的时间,这不可避免的会造成较大的定位延迟。2.信标信号的持续时间过长导致的延迟。要获得位置信息,目标用户需要检测到多个信标信号以收集足够多的时间信息用于定位。在现有的异步定位技术中,信标信号需要借助声空通信技术传输一定的辅助信息,而现有的声空通信技术在动态信道条件下进行可靠传输时,传输速率过低,因而在传输辅助信息时需要较长时间,这就导致信标信号的持续时间过长,在收集足够多的信标信号用于定位时延迟大。为了解决室内定位中的高延迟这一技术挑战,本文在基于智能移动终端声音传感器的室内定位方面进行了深入研究。在针对智能终端的声空信道场景下,提出了基于Chirp信号的归一化信号检测算法与宽正交调制方法作为理论基础,在声空通信应用中,对宽正交调制理论进行了系统验证;在室内定位中,则灵活地利用宽正交调制,从减少搜集多个信标的持续时间和压缩单个信标持续时间方面减少定位中的延迟。具体的,本文的创新点如下:1.为了克服动态信道下信标信号检测不可靠的技术问题,提出了归一化信号检测算法,归一化信号检测算法将信号检测中相关运算后的每一个结果与前面一定数量样本的均值相除,以应对动态信道下的远近效应、设备多样性等问题。针对减少信标持续时间过长的技术问题,则提出了宽正交条件以及宽正交调制方法。当两信号互相关与两信号自相关的较大者间的比值小于一定的阈值时,两信号可视为满足宽正交。上述归一化信号检测算法与宽正交调制是声空通信、以及依赖于声空通信的室内定位的理论基础。2.提出了一个基于宽正交调制的远距离、高速率声空通信技术,克服了常规扩频调制通信技术无法同时实现远距离与高速率两大特性的技术难题。在该通信技术中还引入了可靠的归一化信号检测算法与轻量化的自适应速率调整机制。实验结果表明基于该通信技术的系统在20 m远距离条件下也能保证3%以下的误码率。3.提出了一个基于正交调度与虚拟锚节点的低延迟室内定位技术,在减小定位延迟的基础上克服了现有的基于基础设备的声定位系统无法适应动态环境变化的问题。在该定位技术中,基于宽正交调制方法提出了正交调度的概念,将满足宽正交条件设计的信标分布于不同的锚节点而并行调度,通过减少信标调度时间来克服定位过程中的高延迟;基于正交调度技术,融合了群智感知技术,提出利用定位场景中静态用户作为虚拟锚节点的定位技术,降低定位系统的部署成本和提升其抵御环境变化的能力。实验结果表明,该定位技术能将定位延迟从2 s降低到0.3 s,且90%动态跟踪误差在0.5 m内。4.提出了一个基于宽正交调制压缩信标持续时间的低延迟室内定位技术,解决了定位延迟的技术挑战,实现了实时跟踪。在该定位技术中,通过基于宽正交的OCSS(Orthogonal Chirp Spread Spectrum)调制技术,以压缩信标信号的持续时间的方式克服定位过程中的高延迟;在该定位技术中,对信标信号进行了改进的设计,加入了多普勒载波;提出了新的融合观测时间与运动状态的粒子滤波器模型,允许在单次调度下更新目标位置,使得对移动目标的动态跟踪成为可能。实验表明该定位技术能将定位延迟降低到0.2 s,且90%跟踪误差在0.49 m内。综上所述,本文针对基于移动终端声音传感器的室内定位技术中的高延迟技术挑战,提出了可靠的归一化信号检测算法、宽正交调制技术作为理论基础,在声空通信技术中对此理论基础进行了系统验证,在室内定位中则利用这一理论基础克服了定位过程中的高延迟。更多还原

【Abstract】 Indoor localization is the key enabled technique for location based services.After years of innovative researches,there are now many available localization approaches.Among these localization techniques,acoustic-based indoor localization methods attract much attention for their cost-effective deployment and high-precision localization accuracy.Developing low-latency acoustic-based indoor localization techniques can not only improve user experience but also enables tracking,which could facilitate lots of interesting applications such as tracking and routing.Therefore,researching on low-latency localization techniques is invaluable.However,in dynamic channels where the transceiver distance,type of devices,and motion states are changeable,current asynchronous localization techniques suffer from high latency in obtaining a location fix.The reasons behind high latency are:1.Latency caused by unreliable signal detection under dynamic channels.The unreliable signal detection could lead to false signal detection,miss detection,or large detection errors and thus are filtered out.Therefore,it would need much time to gathering enough beacons for localization,leading to high localization latency.The near-far effect and device diversity problem are the major cause for the unreliable signal detection.2.Latency caused by the time duration of the beacon messages.To obtain a location fix,a target must capture multiple beacon messages.In state-of-the-art asynchronous localization schemes,the beacon messages needs to encode some compulsory information for localization.Due to the low transmission rate,it cause a long time to deliver those compulsory information,making the duration of the beacon messages longer and leading to high latency in localization.In conquering the above high-latency problem in asynchronous localization techniques,we have made in-depth study.We have proposed a normalized signal detection algorithm and a loose orthogonal modulation technique.We verify the above loose orthogonal modulation in a communication system.In the localization techniques,we flexibly utilize the loose orthogonality condition to reduce localization latency.Specifically,by reducing the time to obtain enough beacon messages and compress the duration of a single beacon message,we successfully reduce the latency in the localization systems.Our work innovates in the following aspects:1.To handle the unreliable signal detection under dynamic channels,we propose a simple yet effective normalization method,which utilizes the mean value of several preceding correlation results to normalize the current one.To reduce the latency,we propose a loose orthogonality modulation technique in which two signals are deemed orthogonal if the ratio between the cross-correlation and auto-correlation is below a certain threshold.The above signal detection algorithm and the loose orthogonal modulation technique lay the foundation for our aerial acoustic communication system and localization system.2.We propose a long-range,high-throughput aerial acoustic communication system based on the loose orthogonal modulation technique.In the aerial acoustic communication system,we utilize the loose orthogonal modulation technique to significantly improve the link rate without compromising the communication range.We incorporate the normalized signal detection method and a light-weight rate adaptation mechanism,making the communication system achieve significantly low BER even under long distances.Experiments results present that our proposed method can achieve less than 3% BER under a distance of20 m.3.We propose an orthogonal scheduling and virtual anchors based localization technique,which reduces the latency problem and overcomes the shortcomings of infrastructurebased localization systems that cannot adapt to environment changes.In the proposed technique,we devise an concept of orthogonal scheduling,by which we schedule the orthogonal beacon messages concurrently on distributed anchors,and thereby reducing the latency in obtain sufficient beacon messages.In addition,we incorporate the crowd-sourcing technique which utilizes the stationary clients in the place-of-interest to help locating other targets,eliminating the deployment efforts and making the localization system resilient to environment dynamics.Results show that the proposed approach can reduce the latency from 2 s to0.3 s and achieve a 90% tracking error of 0.5 m.4.We propose a loose orthogonal modulation based localization technique,overcoming the high latency problem and achieving real-time tracking.In this localization technique,we propose a orthogonal chirp spread spectrum modulation scheme which compresses the time duration of the beacon messages and thus reduce the localization latency.In the proposed localization technique,we make several improves.We insert multiple Doppler carriers in the beacon message;we propose a new particle filter model to fuse the observed timestamps and motion states which allows location updates under one scheduling,making it feasible for mobile target tracking.Results reveal a minimal latency of 0.2 s,and a 90% tracking error of 0.49 m.All in all,we have proposed a normalized signal detection algorithm and a loose orthogonal modulation technique for acoustic-enabled mobile devices based low latency indoor localization methods.We verify the above algorithms in an aerial acoustic communication system and utilize these algorithm to reduce the latency in the localization systems.更多还原