Diligence Eagle Optimization Protocol for Secure Routing (DEOPSR) in Cloud-Based Wireless Sensor Network

Abstract

Wireless sensor networks (WSN) are essential for various applications that collect and analyze data from the physical environment. Cloud computing particularly provides a benchmarks method for distributed resource sharing. The concept of “Sensor Clouds” has emerged due to extending the paradigm of cloud computing to include sharing the sensed data through WSN. Data collected by these many sensor-based applications is enormous. Routing is significant in cloud-based WSN (CWSN). Poor routing protocol affects the performance of CWSN severely. In this paper, Diligence Eagle Optimization Protocol for Secure Routing (DEOPSR) is proposed for CWSN. DEOPSR is inspired from natural foraging characteristics of eagle. DEOPSR hold two different types of searching to send data to destination which are local search and global search. Parameter adaption schemes present in DEOPSR assist in selecting the best route and avoiding the selfish nodes in routing. In NS3, the performance of DEOPSR was examined using performance measures prevalently used in computer networks. DEOPSR outperforms the present routing protocols, according to the data.

Data poisoning attacks detection based on the convergence of the Peter and Clark algorithm and Bayesian adversarial federated learning

“…5 Early data poisoning research used manual methods to create fake users, who typically perform suboptimally in attacks even when they have access to input data as well as are aware of the recommendation system. 6 For instance, a random attack selects filler items at random for every fake user as well as rates filler items according to the normal distribution of all rating information. The average attack chooses filler items identically to the random attack.…”

“…There have not been many studies on manipulating ML methods of energy consumption data, despite adversarial ML, and attacks on ML methods, being a developing research field 5 . Early data poisoning research used manual methods to create fake users, who typically perform suboptimally in attacks even when they have access to input data as well as are aware of the recommendation system 6 . For instance, a random attack selects filler items at random for every fake user as well as rates filler items according to the normal distribution of all rating information.…”