车联网与穿戴网融合网络的数据可靠性技术研究

As the rapid development of information technology, the research of vehicle networking and body sensor networks in their respective areas can not meet the growing demand for intelligent life of our people. The integration of vehicle networking and body sensor networks will enable new application areas. However, the integration of the two networks will be fraught with whole new business challenges in communications interference, network attacks, data collision and other areas which affect data reliability. In view of this, based on network information theory, computer science, probability theory and other basic theory, this project first investigates formal verification techniques based on interval temporal logic for the correctness verification of protocols and applications to ensure the data reliability in the logical level; then investigates efficient frequency hopping algorithm to deal with the communication interference and constructs dynamic priority-based data distribution and dynamic, adaptive channel access mechanisms to ensure the reliable data forwarding in the convergence network of vehicle networking and body sensor networks through the underlying communication protocol; finally constructs fault detection and data reconstruction methods based on Bayesian network and Hidden Markov model for detecting and rebuilding faults and data loss to further ensure the data reliability. These techniques can lay the necessary theoretical foundation for the data reliability research in the integration of vehicle networking and body sensor networks, and provide theoretical and technical support for its extensive applications in intelligent transportation, intelligent medical and so on.

随着信息技术的高速发展，车联网和穿戴网在各自领域的研究已不能满足人们日益增长的对智能生活的需求。车联网和穿戴网的融合将开辟全新的应用领域。但是，两网融合在通信干扰、网络攻击、数据碰撞等影响数据可靠性的领域将面临全新的技术挑战。鉴于此，本项目基于网络信息论、计算机科学、概率论等基础理论，首先研究基于区间时序逻辑的形式化验证技术进行各类协议和应用程序的正确性验证，从逻辑层面上保障数据的可靠性；其次，研究跳频等技术应对通信干扰，构建基于动态优先级的数据分发和动态的、自适应的信道接入机制保障数据通过底层的通信协议在车联网和穿戴网融合网络中的可靠转发；最后，构建基于贝叶斯网和隐马尔科夫模型的错误检测和数据修复方法对错误和丢失数据进行检测和修复，进一步确保数据的可靠性；这些技术可为车联网和穿戴网融合的数据可靠性的研究打下必要的理论基础，并为其在智能交通、智慧医疗等方面的广泛应用提供理论和技术支持。

本项目研究基于区间时序逻辑的演绎推理和模型检测两种形式化验技术进行协议和应用程序的正确性验证，研究车联穿戴网中通信干扰应对技术和数据分发及信道接入技术，研究基于距离的错误检测方法，基于隐马尔可夫模型、贝叶斯网的错误检测和数据修复方法。 取得的研究成果包括：（1）提出了一种基于动态贝叶斯网络模型的智能联网车辆数据错误检测与修复方案；（2）提出了基于动态贝叶斯网络模型和时空相关性组合模型的两种异常检测及数据恢复算法，进而给出了最优阈值的性能分析计算方法；（3）提出了一种基于贝叶斯组合模型的生理指标预测机制和基于贝叶斯网络模型的数据修复机制；（4）构建了采用长短期记忆网络（LSTM）对合法用户的步态数据进行端到端的身份认证和采用支持向量域描述（SVDD）来区分合法用户及非法用户的用户细粒度认证模型；（5）提出了一种利用CAN总线上的电压信号的物理特性的基于局部离群因子（LOF）的入侵检测方法；（6）构建了基于卷积神经网络（CNN）和支持向量域描述（SVDD）组合汽车驾驶员指纹识别模型；（7）构建了一种基于贝叶斯推断的信息汇总机制和基于区块链的车联网信任管理模型。