EdgeStream framework cuts IoT latency by up to 92.8% with edge-based streaming

The rapid explosion of IoT devices generates massive continuous data streams that demand real-time processing, but traditional cloud-centric approaches suffer from network delays and high bandwidth costs. To address this, researchers from the team of Atul, Varun Shukla, Vivek Shukla, and Mehul Kumar Das have developed EdgeStream—a lightweight edge-based framework for secure, low-latency streaming analytics. The architecture integrates edge nodes for local processing with a lightweight cloud backend for orchestration and persistent storage, using MQTT as the communication protocol to minimize overhead. EdgeStream also incorporates distributed anomaly detection directly at the edge, enabling immediate response to irregular patterns without round-trips to the cloud. Analytical models for latency, throughput, and bandwidth were developed to benchmark performance.

Experimental results compared EdgeStream against traditional cloud-only systems, demonstrating up to a 92.8% reduction in end-to-end latency while maintaining high throughput and bandwidth efficiency. The framework is especially suited for applications like industrial monitoring, smart cities, and healthcare, where milliseconds matter and network resources are constrained. By moving computation closer to data sources, EdgeStream ensures both security—since sensitive data stays local—and scalability. This unpublished preprint (arXiv:2606.14712) signals a significant step toward practical edge computing for real-time IoT analytics, though real-world deployment and further validation are still needed.