Research And Design Of Secure Storage And Retrieval System For Medical Big Data

With the improvement of living standard, people pay more attention to personal health. At present, there are many mobile wearable medical devices appearing based on intelligent system, which can collect various types of medical data such as blood pressure, blood oxygen, ECG, blood sugar, body fat and upload them to the back-end server.Facing multiple types of massive medical data, one traditional approach is to build the distributed platform Hadoop and store the files on the HDFS directly. In the period of retrieval, access the NameNode through the RPC protocol to obtain the meta-data and DataNode addresses of the file, and then judge whether the file is the one needed according to the meta-data, finally read the corresponding DataNode data.The drawbacks of this method include: the retrieval does a lot of invalid file matching work; remote access NameNode each time, resulting in redundant data interaction and possible time delay; store plaintext files directly, causing the potential safety hazard that medical data information will be leaked once the server is invaded.In view of this, a new storage model is designed in this paper. The model includes: classify multiple types of massive medical data in advance and the classification information is written to the JSON configuration file to save the extra file match actions; After a retrieval job is completed successfully, the retrieval condition and the returned DataNode addresses are stored in a memory hash table and it will be updated when files are changed, in order to avoid frequent remote interactions and improve the retrieval speed; Draw lessons from the latest achievements of cryptography on cipher text retrieval, symmetric searchable encryption algorithm, used in the data storage model in order to strengthen the security of medical data on the server.Based on the above data storage model, this paper designs and implements a secure storage and retrieval system for medical big data. It includes four modules: data classification module, data security module,data storage module and data search module. In the data classification module, multiple types of massive medical data uploaded to the server will be stored in the different positions on the HDFS according to the relevant conditions. The conditions mean keywords to retrieve data later,such as the type of medical data; in the data security module, the security of data will be ensured by two means. The first layer is access control layer, the data access accounts will be authenticated to get the permissions of read, write and execution files, preventing the risk of irregular operations and ultra vires behaviors. The second layer is data encryption layer, using symmetrical searchable encryption algorithm. The files and retrieval conditions will be encrypted using the same key before they are uploaded to the server, ensuring that even administrators on the server side cannot view the content of the medical data files; in the data storage module, Hadoop distributed file system is used, it has high reliability and high fault tolerance, ensuring that the medical data will not be easily lost through the way of distributed storage to save multiple copies of the source files; in the data retrieval module, a Hash table in memory is maintained. After a successful retrieval is completed, retrieval conditions and DataNode addresses of the retrieval files will be stored in the Hash table. Each of the four modules completes single task to reduce the coupling between modules, and improves the scalability of the system.When the system is completed, the security of the system is evaluated according to the encryption algorithms used. Then we store multiple types of massive medical data in a single machine and this system to do several retrieval works respectively. The retrieval efficiency of this system is higher from the consumption time of the retrieval operations.In the end of the paper, we summarize the work, elaborate the achievements of this topic, point out the deficiency existing in the system and discuss the development of the secure storage and retrieval system for medical big data.