Adaptive multidimensional trust-based recommendation model for peer to peer applications

Abstract

In today’s world, the widespread utilization of services such as Nearby Share, Near Field Communication (NFC), and Wi-Fi Direct for deploying various applications has significantly bolstered the demand for reliable and secure distributed ad-hoc peer-to-peer networks. Yet, ensuring the trustworthiness of participating nodes remains a significant challenge. Trust among nodes plays a pivotal role in collaborative network applications, especially in environments like Mobile Ad-hoc Networks and VANET (Vehicular Ad-hoc Networks). Evaluating the trustworthiness of nodes is essential for promptly identifying misleading entities, thereby preemptively preventing their involvement in ongoing transactions. Attributes or characteristics exhibited by nodes, such as honesty, selfishness, or malicious behavior, serve as key factors in trust computation. The effectiveness of trust evaluation directly influences the encouragement of honest nodes and the deterrence of malicious ones, thereby nurturing a healthy and competitive network ecosystem. Recognizing the dynamic nature of network environments, trust computation methods must be adaptable and diverse. The adaptive multidimensional trust (AMT) model introduced in this article goes beyond simple reputation assessment. It offers three distinct methods such as Direct Trust ( Direct Trust ) \left({{\rm{Direct}}}_{{\rm{Trust}}}) , multiple security parameters, identification of qualified recommenders, which got selected dynamically as per change in trust ratings of peers. AMT advocates for an incentive-driven approach to identify legitimate peers, monitoring gradual increases in their performance ratings, whereas, spikes in performance alert to potential colluding peers or nodes displaying erratic behavior. This article evaluates the effectiveness of the AMT through a case study focused on an E-commerce application. It scrutinizes the model’s performance across different percentages of malicious nodes within the network, providing a thorough analysis and discussion of the results based on the trust value of malicious and benign peers and efficiency by selecting genuine service for transaction.