SCMBQA: Design of a Customised SCM-Aware Sidechaining Model for QoS Enhancement under Attack Scenarios

Abstract

Storing& processing data for supply chain management (SCM) systems requires design of high-security and quality of service (QoS) aware models. These modelsassist in improving traceability performance of SCM deployments via integration of transparent & distributed mechanisms. A wide variety of security models are proposed by researchers to perform these tasks, and it is observed that blockchain-based SCM implementations outperform other models in terms of security & QoS metrics.But most of these implementationsare general-purpose and do not incorporate SCM-specific consensus & mining rules. It is also observed that, mining speed& throughput performance of these blockchain-based implementations reduces exponentially w.r.t. number of SCM transactions. To resolve these issues, this paper discusses design of a novel Proof-of-Supply Chain (PoSC) based consensus model, which is specifically designed for sidechain based SCM deployments. The PoSC consensus model is used for high-efficiency SCM-based data storage and communication scenarios. The proposed PoSC consensus model is capable of resisting selfish mining, time jacking, and sybil attacks, which are targeted towards SCM deployments. The model uses temporal performance metrics of miner nodes, and combines them with relationship graphs to form an SCM miner rank. Based on this rank, miner nodes are selected, and their consensus responses are recorded. These responses are processed using an augmented deep learning model, that is trained over 8 different SCM implementations via machine learning. After successful mining, responses obtained from these miners are used to incrementally train the machine learning model which assists in continuous performance improvement. The SCMBQA model was tested on milk supply chain, agriculture supply chain, and electronic supply chain applications, in terms of computational speed, throughput, energy requirement, retrieval & verification delay, and storage requirements. It was observed that the proposed PoSC consensus was capable of improving the computational speed by 8.5%, reduce energy consumption by 4.9%, improve throughput by 9.6%, and reduce storage costs by 15.4% when compared with standard blockchain-based SCM consensus models. This is because the proposed model deploys an intelligent sidechaining approach, that is capable of optimizing number of generated sidechains via temporal QoS & security performance metrics. Due to use of smaller chain lengths, the proposed model is capable of integrating privacy-aware & secure approaches depending upon different SCM stages. Thus, distributor-level security models are different than retailer-level security models, which assists in context-sensitive block deployments. Due to use of PoSC, the proposed model was observed to be 99.5% resilient against internal and external attacks, which makes it useful for real-time SCM deployments.