A Novel Architecture For Network Coded Electronic Health Record Storage System

The use of network coding for large scale content distribution improves download time. This is demonstrated in this work by the use of network coded Electronic Health Record Storage System (EHR-SS). An architecture of 4-layer to build the EHR-SS is designed. The application integrates the data captured for the patient from three modules namely administrative data, medical records of consultation and reports of medical tests. The lower layer is the data capturing layer using RFID reader. The data is captured in the lower level from different nodes. The data is combined with some linear coefficients using linear network coding. At the lower level the data from different tags are combined and stored and at the level 2 coding combines the data from multiple readers and a corresponding encoding vector is generated. This network coding is done at the server node through small mat lab net-cod interface software. While accessing the stored data, the user data has the data type represented in the form of decoding vector. For storing and retrieval the primary key is the patient id. The results obtained were observed with a reduction of download time of about 12% for our case study set up.


Introduction
The fundamental idea of Network Coding spreads its potential in various network Nowadays machine generated data exceeds human generated data and we need storage systems of the order of Exabyte's. When we consider distribution of data in wireless networks, there exists challenges like wireless data rates, link failures and packet loss probability. Thus we need to develop strategies by deployment and integration of newer technologies. The desirable performance metrics includes rebuild time, read/write bandwidth and storage efficiency. The major pulling factor is the tradeoff between reliability and redundancy. Intelligent architectures are required to achieve this and one such effort is this work by focusing on the metric rebuild time.
We make use of content based network coding for a selected scenario of EHR, which results in better storage and retrieval of contents. The rest of the paper is organized as follows. Section 2 reveals the related works and the basic principle used in our work.
In section 3 we propose a new architecture for the design of network coded EHR-SS.

Related Works
Network Coding allows more intelligence at the nodes to perform simple computation (encoding). The data packets are combined and stored for distributed storage. Also the profit of network coding is achieved using linear transformations.  To illustrate how network code improves the propagation of information without a global coordinated scheduler we consider the following (simple) example. In Figure 6.1 assume that Node A has received from the source packets 1 and 2. If network coding is not used, then, Node B can download either packet 1 or packet 2 from A with the same probability.
At the same time that Node B downloads a packet from A, Node C independently downloads packet 1. If Node B decides to receive packet 1 from A, then both Nodes B and C will have the same packet 1 and, the link between them cannot be used.
If network coding is used, Node B will download a linear combination of packets 1 and 2 from A, which in turn can be used with Node C. Obviously, Node B could have downloaded packet 2 from A and then use efficiently the link with C. We propose a architecture of 4-layer to build the HER-SS as shown in Figure   3.1. The lower layer is the data capturing layer. This uses RFID passive tags. The captured details are uploading on the clients in the 2 nd layer. The networked clients are connected in this layer and they upload the data to the server in the 3 rd layer. The network coded details exists in both 2 nd and 3 rd layer. The data cloud is optional and is constructs the 4 th layer of the architecture.

Figure 3.1 Architecture of EHR-SS
The network coding strategy used is as described in and the notations used are described in Table 3.1.

RFID based Data Capture
The RFID system consists of a reader, tag and the host system. The reader and tag communicate through a RF signal link.    The strategical approach followed is described here. The data is captured through a .net application and the solution pushes the data to be combined into a network coding middleware. The middleware operation is performed in the matlab environment.
The export and downcasting are the operations building the intraoperability.
The network coding middleware's abstracted view is shown in Figure 3

Client Server Model
The RFID readers are connected with the clients by Wi-Fi network model.
The client support different mobile RFID readers. From the client the environment uses the local area network to connect the data to server. The hospital main server is visualized to get connected with many clients in the hospital environment.

EHR Scenario
The idea of our proposal addresses a Pervasive Hospital environment from where the data is uploaded to the cloud server. Also the data can be downloaded from the cloud to the user community. The overall architecture is modelled as a 4-layer set up and is depicted in  informations are coded and tested. The data storage and retrieval of network coded output is done with a local application server.

Output Samples
In the first stage of experiment, the application is deployed in the handheld reader. The .net based user interface as shown in Figure 5.1 allows the user to interact through the device. Synchronization of the device is essential and the output sample in Figure 5.2 reveals that. This invokes the application at the reader. The patient id is read as per sample in Figure 5

Conclusion
A simple linear network coding application is verified through our case study.
The benefits of our design are (i) the server application is ignorant about the source of the contents collected at a specific time stamp (ii) while retrieving the data, the hand held device requests are very simple without specifying the details of their module; this saves the transmission cost (iii) In case of malfunction of any of the clients, the information can be retained with the history of encoding variables and the neighbour clients.