基于低成本无线传感器网络的实时环境噪声数据获取问题研究

With the rapid development of modern society, the negative effects of noise pollution on public health, animal and plants growth are exacerbated, that seriously hinder the sustainable development of ecological environment. To obtain detailed noise distribution information, many countries have launched the noise mapping strategic action. The academics have made great contributions for real-time noise data acquisition in terms of low cost sensor node design, system architecture and binary noise determination. However, there are still many open research issues that need to be addressed, such as high sensing energy consumption, resource constraint on sensor nodes and lossy wireless networks. This project focuses on the challenges of environmental sampling, processing and wireless transmission during noise data acquisition with low cost wireless sensor networks. Firstly, a joint design of passive event trigger circuit and energy-aware task scheduling is presented for adaptive noise sampling. Secondly, lightweight noise feature extraction and classifier based on approximate computing are proposed for noise sources recognition on resource-constrained embedded platforms. Finally, a wireless communication approach that combines with concurrent transmission and network coding mechanism is proposed to improve the reliability of noise data delivery. The research project has great significance for the fundamental research achievement of wireless sensor network to practical application in ecological environment protection.

随着现代社会的快速发展，噪声污染问题凸显，其持续危害人类健康、畜禽发育和农作物生长，严重阻碍生态环境的可持续发展。为获得详细噪声分布信息，世界各国陆续启动噪声地图战略计划。针对实时噪声数据获取问题，学术界在低成本传感器节点设计、系统架构和二元噪声判断等方面取得了重大进展。但仍存在能耗过高、资源受限和无线链路不稳定等尚未解决的核心科学问题。本课题围绕基于低成本无线传感器网络噪声数据获取过程中必经的环境感知采样、计算处理和无线传输三个步骤开展系统性研究。首先，采用基于无源器件触发电路设计和能量意识任务调度的软硬件协同方法，达到环境自适应噪声采样；然后，提出基于近似计算思想的噪声特征提取与分类器设计，实现资源受限平台的轻量级多元噪声识别；最后，利用并发传输和网络编码相结合的通信机制，提高噪声信息无线传输可靠性。本项目研究对无线传感器网络基础研究成果走向生态环境保护应用具有重要意义。

随着现代社会的不断发展，噪声污染问题凸显，其持续危害人类健康、畜禽发育和农作物生长，严重阻碍生态环境的可持续发展。为治理环境噪声污染，世界各国陆续启动噪声地图战略计划。我国在“十四五规划”中，明确提出“加强环境噪声污染治理”，这也是噪声污染问题首次纳入国民经济和社会发展规划。. 通过无线声音传感器网络对环境噪声进行监测，是一种高效智能的解决方案。目前学术界在低成本传感器节点设计、系统架构和二元噪声判断等方面取得了重大进展。但仍存在能耗过高、资源受限和无线链路不稳定等尚未解决的核心科学问题。本课题围绕基于低成本无线传感器网络噪声数据获取过程中必经的环境感知采样、计算处理和无线传输三个步骤开展系统性研究。重要结果总结如下：.(1) 在环境感知采用方面，提出了两种基于被动感知的按需工作声音传感器节点架构。该方案可以实现空闲期的近零能量消耗，并在关键声音信号来临时快速唤醒进行精细化感知。.(2) 在计算处理方面，通过降维和近似计算思想，以及代码优化，设计了能够在资源受限平台下运行的轻量级多元噪声识别算法，从而实现传感计算与边缘计算的协同噪声识别。.(3) 在无线传输方面，研发了无需基础设施支持的传感网络测试平台ChirpBox，并且进行硬件平台及软件代码开源。对基于LoRa的户外广域网络进行长期网络连通性和链路质量分析，并得出网络性能与环境因素（温度、湿度和风速）的影响关系。提出基于并发传输的Harmony协议，抵抗网络外部射频干扰，实现高可靠无线通信。. 本课题研究以噪声污染问题及噪声可视化为需求牵引，突破低成本无线传感器网络噪声数据采集所面临的技术瓶颈，解决其尚未大规模应用背后的核心科学问题，促使无线传感器网络基础理论研究成果和技术成果可以更好更快的走向生态环境保护应用领域，并服务于国家生态环境保护和智慧城市数字化战略需求。