Research On Secure Outsourcing Computing For The Internet Of Thing

Internet of Things(IoT)has developed into an emerging strategic industry at the national level,it has become another wave of development after computers and the Internet.By utilizing Internet of Things technology,various terminal devices send the collected information such as sound,light,biology and location to the network,so as to realize the connection between things and people,and further realize the intelligent perception,recognition and management of goods and information in the processing process.At present,the Internet of Things technology is mainly used in smart home,smart industry,smart logistics,smart power,smart medical and other fields.However,due to the large amount of data and the diversity of data types in the Internet of Things,some nodes in the Internet of Things,may not be able to bear the heavy data processing tasks.Outsourcing computation just solves this problem.With the technology,resource-constrained users can outsource the planned computing tasks to the cloud servers to reduce their own computing burden while obtaining the results after data processing.When dealing with massive data in the Internet of Things,outsourcing computing technology has many advantages,it also brings security challenges.First of all,terminal devices in the Internet of Things are often directly connected to people or things themselves,which inevitably involves private information,such as trade secrets,patient identity information,health information,etc.Secondly,the cloud servers are not fully trusted.In other words,during the outsourcing process,the sensitive information in the data may be stolen.In addition,software errors or malicious attacks may cause errors in the computation results returned by the cloud servers.Aiming at the specific application of power and medical in the Internet of Things,the thesis designs privacy-preserving outsourcing schemes for harmonic estimation and medical image diagnosis in the power system using outsourcing computing technology,it includes:(1)For the harmonic estimation in the industrial internet of things,the first secure outsourcing computation scheme for harmonic estimation in power system is proposed.This scheme protects the sensitive data in the outsourcing algorithm by designing a new matrix transformation method.With the help of cloud servers,this scheme can complete the harmonic estimation algorithm in resource-constrained IoT devices.At the same time,this method greatly saves the computing resources required for the implementation on the client side.This experiment proves that the cost of the proposed scheme at the user end is far less than that of the original harmonic estimation calculation itself.At the same time,the correctness,verifiability and efficiency of the algorithm are also analyzed and proved through detailed theoretical description.(2)For the task of intelligent medical traditional Chinese medicine image classification,a privacy-preserving outsourcing algorithm for medical image diagnosis based on convolution neural network is proposed.The algorithm outsources the training process of medical image diagnosis based on convolution neural network model to the cloud service platform,the obtained model can be used for disease diagnosis and speculation of other users in the Internet of Things.The scheme blinds information in the system with a series of lightweight cryptography primitives,which means that the cloud service platform cannot obtain user information,the model information of the convolution neural network model used for medical image diagnosis is also protected.The efficiency of this scheme has also been proved by the experiment of pathological section staining.