A Blockchain-Enabled Iot Framework With Rule-Weighted Quality Scoring For Transparent And Tamper-Evident Agri-Food Supply Chain Traceability

Abstract

Agri-food supply chains connect farmers, processors, distributors, retailers and consumers across multiple perishable-sensitive stages, and their performance directly shapes food safety, quality retention and market trust. Digital transformation initiatives increasingly combine Internet of Things (IoT) sensing with distributed ledger technology to make these multi-echelon chains more transparent, traceable and resilient. Despite this potential, existing agri-food chains still suffer from fragmented, unverifiable quality records, delayed anomaly detection and centralized data stores that remain vulnerable to tampering and single points of failure. Prior blockchain-oriented proposals in this domain are frequently conceptual, rely on subjective expert-elicitation methods, or omit reproducible, empirically benchmarked implementations of the traceability and quality-assessment layers. This study designs and implements AgriFoodBlockchain, a six-layer framework that couples IoT sensor acquisition with a Proof-of-Work blockchain ledger and a rule-weighted quality-scoring engine. A dataset of 120 IoT observations covering 30 agricultural batches, six product types, four supply-chain stages, eight locations, twenty farmers and forty-five IoT devices was processed end-to-end through the framework, with each record scored, graded and permanently anchored into a hash-linked chain. Algorithm 1 implements a rule-weighted quality-scoring model that aggregates temperature, humidity, pH, carbon dioxide, dissolved-oxygen and ethylene readings into a single 0–100 index using empirically assigned feature weights. Algorithm 2 implements block creation and chain validation, combining previous-hash linkage, a three-leading-zero Proof-of-Work condition and batch-index lookup to guarantee tamper-evident storage. The framework processed all 120 records with 100% data completeness, producing a 121-block chain that passed full integrity and linkage validation. The mean quality score across all records was 78.77 (SD = 14.35), with 48.3% of records graded EXCELLENT and 42.5% GOOD, while Wheat (94.50) and Storage-stage records (82.53) achieved the highest mean scores. Benchmarking over 1,000 repetitions showed an average in-memory batch-index lookup latency of 302 nanoseconds and an observed throughput of 16.9 records per second, with zero POOR or CRITICAL anomalies detected. These results indicate that a lightweight, interpretable rule-weighted scoring model combined with a Proof-of-Work ledger can deliver complete, verifiable and low-latency traceability without requiring a trained machine-learning classifier. Future work will extend the framework toward multi-node consensus, adaptive feature weighting and smart-contract-based automated recall mechanisms.