PoPEA: an energy and latency-optimized proof-of-processing framework for trusted end–edge–cloud healthcare analytics

Abstract

In smart healthcare, continuous patient monitoring relies on streaming analytics across end, edge, and cloud layers. While edge computing enables low-latency analytics, ensuring that analytics are correctly performed by potentially untrusted nodes remains a significant challenge. We propose PoPEA-BC, a blockchain-enabled Proof-of-Processing framework combining Trusted Execution Environments (TEEs), succinct zero-knowledge proofs (zk-proofs), and distributed ledger technology (DLT) to cryptographically verify that prescribed analytics were executed as intended and immutably record proofs for auditability. PoPEA-BC introduces an operator-level proof library for common healthcare stream analytics, including ECG feature extraction and arrhythmia detection, and a hybrid MILP/Reinforcement Learning-based scheduler that optimizes latency, energy consumption, and proof generation overhead under intermittent connectivity. Blockchain provides a tamper-proof, decentralized verification layer for storing computation proofs, enabling post hoc validation by authorized healthcare stakeholders without relying on a single trusted authority. Evaluation on real ECG and SpO2 anomaly detection workloads demonstrates that PoPEA-BC achieves verifiable processing with <8% additional latency, reduces energy consumption by up to 24% compared with TEE-only baselines, and remains resilient against Byzantine edge nodes. These results highlight PoPEA-BC's potential as a foundational layer for trustworthy, auditable, and efficient healthcare analytics in distributed IoT environments. Although PoPEA-BC demonstrates strong performance under controlled settings, its evaluation is limited to publicly available physiological datasets and small-scale consortium blockchain deployments. Future work will explore large-scale real-world hospital integration and lightweight proof optimizations for ultra-low-power devices.