Ridezy derives driver credibility in real time from edge AI and IoT sensors and anchors hashes and reputation updates on Polygon, outperforming rating-based, AI-only, and blockchain-only baselines in behavioural fidelity and trust guarantees

It proposes a sensor-driven credibility framework that transforms real-time behavioural evidence into a cryptographically verifiable trust primitive, replacing reputation mechanisms that rely on subjective post-ride ratings. Prior studies explored blockchain-based ride-sharing, AI-driven driver monitoring, and IoT-enabled transportation separately but did not provide an integrated trust framework that continuously derives credibility from objectively verifiable driving behaviour while maintaining scalability and low operational cost; this work integrates the three into a single architecture. The claim is supported by the architecture design and a comparative evaluation: Ridezy is compared against three representative trust models (traditional rating-based, AI-only, and blockchain-only) on behavioural detection performance, end-to-end latency, cost efficiency, throughput, reputation stability, tamper resistance, and component-wise ablation, with results reporting higher behavioural fidelity and stronger trust guarantees for the hybrid architecture.

It adopts a hybrid architecture that processes behavioural summaries off-chain and anchors only cryptographic hashes, credibility updates, and payment records on Polygon smart contracts, achieving tamper-evident trust with reduced blockchain overhead. Unlike storing high-frequency telemetry directly on-chain, this design keeps high-frequency data off-chain and puts only verifiable summaries and records on-chain, preserving tamper evidence while lowering per-ride cost. The evaluation reports that the architecture maintains low per-ride blockchain cost and scalable operation through batched on-chain commitments; cost efficiency and throughput are among the evaluation dimensions.

Through component-wise ablation and multi-dimensional comparison, it indicates that combining edge AI, IoT telemetry, and blockchain-based trust anchoring outperforms any single-component path. The AI-only and blockchain-only architectures are included as controls, making the gain of the hybrid approach over single-technology routes a testable comparison rather than a design claim alone. Evidence comes from component-wise ablation studies and comparison against the three baselines, covering behavioural detection performance, latency, cost, throughput, reputation stability, and tamper resistance.

The work targets continuous assessment of driver credibility in ride-sharing ecosystems, applicable to platforms with in-vehicle or mobile sensing, edge compute, and blockchain access; its value lies in turning real-time behavioural evidence into a cryptographically verifiable trust primitive, providing a basis for later exploration in real fleets, cross-platform reputation recognition, and regulatory audit interfaces. For a reader, this means reputation can be understood as independently verifiable behavioural records rather than passenger scores alone.

The currently visible text is an incomplete read and does not include specific experimental values, datasets, sample sizes, statistical tests, or figures, so the magnitude of claims such as 'higher behavioural fidelity', 'stronger trust guarantees', and 'low per-ride cost' cannot be verified. A careful reader would still watch how behavioural detection performs on real roads and under adverse sensing conditions, how batched on-chain commitments affect latency and throughput in practice, and how reputation stability holds under long-term operation and adversarial manipulation.