Fog Node Selection Algorithm Based on the Similarity of Component Degrees

The FC (Fog Computing) model is discussed to implement the IoT (Internet of Things). Here, it is critical to protect fog nodes and devices from malicious accesses since both are only supported with poor computation resources. In order to protect devices and fog nodes from malicious accesses, the CBAC (Capability-Based Access Control) model is used. However, a node can get the data which the node is not allowed to get via other nodes. Here, illegal information flow occurs. The sensor data are partially processed at fog nodes and only processed data are exchanged among nodes. We consider a concept of component degree showing how much sensor data are in the output data from nodes. We propose the RR (Request Rejection) and RRS (RR with the fog node Selection algorithm) protocols where requests occurring illegal information flow are rejected in our previous studies. In the RRS protocol, nodes are selected to reduce the requests rejected. In this paper, we propose a selection algorithm based on the cosine similarity. Here, nodes are selected based on the cosine similarity between a vector which shows devices where a node is allowed to get data and component degrees of the output data. The ratio of requests rejected in the RRSC (RRS based on the Cosine similarity) protocol is smaller than the RR protocol but larger than the RRS protocol. However, it is expected that component degrees can be reduced by using the dimensionality reduction in the RRSC protocol.