A Smart LoRaWAN-based Diabetic Monitoring System using Dual Coordinate Descent SVM for Data Transmission Optimization

Due to the exploding number of remote healthcare
applications, the need for reliable and energy-efficient data
transmission for continuous diabetic monitoring is also on the
rise. However, the conventional LoRaWAN based systems have
static or rule-based scheduling mechanisms which lead to
excessive energy consumption, network congestion and delayed
detection of critical health events. The goal of this study is to
create an intelligent information communication framework
based on adaptation that is efficient in terms of data transmission
and safe from a clinical standpoint. This paper proposes a smart
Diabetic monitoring system using LoRaWAN with Dual
Coordinate Descent Support Vector Machine (DCD-SVM) for
Risk Aware and QoS based transmission control. The model
dynamically classifies patient states and dynamically alters
spreading factor, transmission power, packet interval and
routing paths. Extensive NS-3/LoRaSim simulations have shown
that the proposed DCD-SVM-AT framework provides a packet
delivery ratio of 98.4%, average end-to-end delay is 140 ms,
energy consumption is 2.1 J and network lifetime is 2730 s. In
addition, it achieves a transmission reduction ratio of 43%,
critical event detection latency of 4.6 seconds, and critical event
missing rate of 1.1%, which is better than Static LoRaWAN,
ADR, QoS-routing, LSTM-AT and XGBoost-AT schemes. These
results confirm that lightweight edge intelligence and the
combination with adaptive LoRaWAN communication is a
promising and powerful method to boost reliability, efficiency
and clinical responsiveness of large-scale diabetic monitoring
systems.