A Multi-Layer ICT Architecture for Reliable Embedded Multi-Sensor Monitoring

Information and Communication Technologies (ICT) are the backbone of modern monitoring infrastructures that interconnect specific, reprezentative and conclusive physical sensors, embedded controllers, wireless links, and cloud services. Designing such systems requires more than component integration: it demands a coherent view of data acquisition, on-node mathematical preprocessing, link-layer reliability, and end-to-end latency under realistic operating conditions. This paper presents a multi-layer ICT architecture for embedded multi-sensor monitoring systems, together with mathematical signal models used to clean, cluster, and fuse measurements before transmission.

The proposed architecture organizes the ICT pipeline into six cooperating layers—physical sensing, microcontroller preprocessing, local field buses, wireless gateway, cloud API, and user dashboard—and evaluates three communication technologies (Wi-Fi, LoRa, BLE) in terms of packet delivery ratio (PDR) and latency. On-node processing combines lightweight filtering with a Gaussian mixture model (GMM) for unsupervised characterization of multi-modal sensor distributions, followed by a fusion stage that reduces reconstruction error relative to single-sensor baselines. Laboratory experiments on a multi-sensor testbed report mean end-to-end availability of 98.9%, Wi-Fi PDR above 99% for small payloads, and fusion RMSE of 0.041 (normalized), with mean Wi-Fi path latency below 65 ms. The contribution is a practically validated ICT blueprint that couples mathematical signal modeling with communication-aware design for reliable embedded monitoring.