1. Random Probit Regressive Decision Forest Classification based IoT aware Content Caching with Healthcare Data.
- Author
-
Sangeetha, R. and Ravi, T. N.
- Subjects
PROBIT analysis ,DECISION trees ,RANDOM forest algorithms ,INTERNET of things ,REGRESSION trees ,CACHE memory ,PLURALITY voting - Abstract
Information-Centric Networking (ICN) is promising system architecture for distributing popular information across the network. Its important feature is the node cache. ICN with caching is a very promising featured network structural design. There are numerous cache nodes distributed in the ICN. Excessive consumption of cache information on a tremendous number of nodes dissipates a large storage space and causes data redundancy, which reduces the cache hit rate. Therefore, an efficient caching deployment approach is required to improve the cache hit rate. A novel RAndom probit regressive Bucklin DEcision Forest classifier (RADEF) technique is introduced in Information-Centric Networking (ICN) for minimal network latency and higher cache hit rate. In RADEF Technique, probit regression and random forest classification processes are carried out for ICN with healthcare patient data. First, the patient information is collected from different IoT devices and registered. Every request to the router node is analyzed in the content storage (CS) by using a random probit regressive Bucklin decision forest classifier. The Bucklin decision forest classifier is an ensemble technique that includes a set of weak learners as decision trees (i.e., probit regression tree) and IoHT data are selected randomly for each decision tree. The Probit regression tree is constructed to analyze the patient healthcare data request search in the content storage (CS) of the router node in the ICN network by using a simple matching coefficient. If the copy is present in the content storage, the particular router nodes are chosen and deliver the content. If the copy is absent, the patient information is stored in the cache. Then, the weak learner's results are combined to make a strong output. Then the votes are generated for each decision tree. The votes of all decision trees are combined to identify the majority votes of data for classification by minimizing the error using the Bucklin voting method. In this way, content catching is effectively performed in ICN with minimum latency and a higher cache hit ratio. Experimental evaluation is carried out on factors such as cache hit rate, network latency, average request length, average response time, server traffic ratio, and hop reduction ratio regarding respect to the number of patient healthcare data. The analyzed results demonstrate the superior performance of our proposed RADEF technique when compared with existing methods. [ABSTRACT FROM AUTHOR]
- Published
- 2024