The rapid advancement of wireless and cyber-physical systems has driven a growing demand for real-time sensor data in cyber environments. While existing solutions attempt to meet these demands, achieving both high-throughput processing and robust data integrity remains a significant challenge. This paper proposes an end-to-end distributed data pipeline and blockchain-enabled framework that integrates online anomaly detection, scalable data aggregation, and secure verification to ensure reliable cyber evolution. An Apache Kafka-based streaming pipeline ingests high-velocity sensor data and employs a dynamic bucketing strategy that finalizes buckets based on data volume, elapsed time, and network gas costs. Once validated, each bucket’s canonical sensor data representation is hashed and committed on-chain for tamper-evident storage. To enhance efficiency and security, the framework supports both Merkle tree and Verkle tree cryptographic data structures for comparative analysis. Implemented on a private Ethereum-like blockchain, our system efficiently handles large-scale sensor ingestion while enabling per-record verification. By integrating real-time anomaly correction, cryptographic proof mechanisms, and on-chain commitments, our solution delivers trustworthy, verifiable sensor streams tailored to the low-latency and high-reliability needs of next-generation systems.

As machine learning models, specifically neural networks, are becoming increasingly popular, there are concerns regarding their trustworthiness, specially in safety-critical applications, e.g. actions of an autonomous vehicle must be safe. There are approaches that can train neural networks where such domain requirements are enforced as constraints, but they either cannot guarantee that the constraint will be satisfied by all possible predictions (even on unseen data) or they are limited in the type of constraints that can be enforced. In this paper, we present an approach to train neural networks which can enforce a wide variety of constraints and guarantee that the constraint is satisfied by all possible predictions. The approach builds on earlier work where learning linear models is formulated as a constraint satisfaction problem (CSP). To make this idea applicable to neural networks, two crucial new elements are added: constraint propagation over the network layers, and weight updates based on a mix of gradient descent and CSP solving. Evaluation on various machine learning tasks demonstrates that our approach is flexible enough to enforce a wide variety of domain constraints and is able to guarantee them in neural networks.