| The extensive adoption of emerging technologies,including NB-IoT,5G,cloud computing,cloud storage,and big data,coupled with the rapid development and widespread use of smart devices,has significantly bolstered IoT applications and innovations.With the escalating number of IoT devices,it has become particularly critical to use mobile terminal software to rapidly connect and manage them efficiently.However,common IoT mobile terminal software available in the market lacks a unified standard for device access and a security mechanism for device authentication and user verification.Moreover,the existing IoT mobile terminal software faces challenges in responding to changes in IoT scenarios and implementing complex services,which ultimately leads to poor user experience.Lastly,most of the current IoT mobile terminal software processes perception decisions in the cloud,resulting in significant transmission delays that fail to meet the high real-time business requirements.This paper aims to address the issues associated with existing IoT smart terminal software and proposes the design and development of a comprehensive IoT mobile terminal software.Firstly,the paper investigates techniques to improve device access compatibility and security.Additionally,it introduces a rule engine-based inference decision mechanism to enhance overall functionality.Moreover,the paper presents the design of neural network models specifically for gesture and speech command recognition,which are successfully deployed on the mobile terminal,enabling intelligent sensing decisions with significantly reduced latency and power consumption.The main contributions and outcomes of this thesis are summarized as follows.1.The overall architecture of the web-based IoT mobile terminal software has been designed.Based on the business logic,the architecture is divided into five modules:user management,scenario management,device control,device access and command recognition.The modules are decoupled to enable functional separation and orderly collaboration.2.A three-way collaborative access authentication mechanism among the cloud,mobile terminal software and the device to be accessed has been designed.The "Thing Model" is utilized in the device access module to standardize device data and facilitate access to different devices.The compatibility and security of device access are ensured through the interaction and collaboration authentication among the three parties.3.A rule engine-based reasoning and decision-making scheme has been proposed to achieve decision control based on scene fluctuations.The rule engine abstracts these simple and independent rules defined in the scenario management module into complex business logic and generates inference networks,continuously matching environment variables with the rules in the rule base to facilitate decision inference and execute commands on the devices.4.Gesture and speech command recognition models based on the mobile terminal have been designed,implemented,trained,and localized for deployment.Edge computing is incorporated into the mobile terminal to address network resource constraints and reduce high latency in the perception and decision-making process of IoT.Test results demonstrate that gesture recognition achieves an accuracy of 0.99 with an average recognition time of 10ms,while speech recognition reaches an accuracy of 0.91 with an average recognition time of 331ms.Command delivery time is under 200ms,ensuring accurate and low-latency device control.5.The development of the generic IoT mobile terminal software has been successfully completed,and the modules have undergone testing in smart home and health monitoring IoT scenarios.The test results indicate that the software exhibits good response speed,frame rate and power consumption performance,providing users with a satisfying experience. |
