Journal of Electronics and Telecommunication System Engineering

Smart Agriculture System for Real-time Plant Disease Detection Using Transfer Learning and Uncertainty-aware Deep Learning

2026-05-26T04:12:53+00:00

Plant diseases are among the most persistent threats to agricultural productivity, responsible for an estimated 20 to 40 percent of global crop losses every year. In most farming communities, especially small-scale and rural ones, disease identification still depends on manual inspection by trained agronomists, a process that is slow, costly, and simply unavailable to the majority of farmers who need it most. By the time visible symptoms are identified and a diagnosis is made, infections have often already spread across a significant portion of the crop. This delay between onset and detection is where the largest share of yield loss occurs, making early and accurate identification not just useful, but critical. The system classifies 38 diseases and healthy states across 14 crop species from live webcam footage or uploaded leaf images, filters out non-leaf and ambiguous inputs automatically, and communicates results both through a browser-based web interface and through a physical LED indicator connected via an Arduino microcontroller. The detection model is built on MobileNetV2, a lightweight convolutional neural network architecture designed specifically for deployment on resource-constrained devices. Rather than training from scratch, the model is initialized from ImageNet-pretrained weights and fine-tuned on the PlantVillage dataset, which contains 54,306 labeled leaf images. Transfer learning in this manner dramatically reduces the training data and compute time required while preserving strong generalization capability. An entropy-based uncertainty filter is layered on top of the classifier so that inputs lacking sufficient confidence, such as non-leaf objects or blurry frames, are rejected rather than misclassified. The system is expected to achieve a validation accuracy of approximately 95.41% across all 38 classes, with per-frame inference latency of 30 to 60 milliseconds on a CPU fast enough to support smooth live detection through the webcam stream. Beyond accuracy, the work aims to demonstrate that a fully functional agricultural AI tool can be built.

Black Box with Integrated Accelerator for Real-Time Multi-Sensor Fusion: A Survey

2026-06-23T11:49:42+00:00

The rapid technological progression of intelligent transportation systems has, in turn, escalated the requirement for dependable, secure, and real-time vehicular data acquisition methods for accident analysis, driver behavior monitoring, and road safety enhancement. Today, automotive black box systems have materially transformed from mere crash data recorders to sophisticated platforms that integrate numerous sensors, wireless communication, and data analytics. This survey article aims to cover a wide range of past research on black box systems in the literature published in recent years. Their quality, diversity, and voluminous character made it necessary to develop robust methodologies for researching structures and to apply these methodologies for their comprehensive analysis, classification, and description in this survey article. Apart from the system architecture, the paper also investigated a broad palette of related research topics, such as sensing technologies, communication methods, data storage tactics, and intelligent processing techniques. The comparative performance review highlights the advantages and limitations of existing solutions using at least five performance indicators: reliability, latency, scalability, security, and automotive suitability. The survey introduces a taxonomy of black box technologies and also pinpoints critical research areas that may be filled by real-time multi-sensor fusion, secure data storage, and automotive-grade hardware compliance. Various topics covered in the article are being implemented in the industry, such as integrating Internet of Things technology, cloud-based data analytics, and intelligent accident detection. Lastly, this paper gives directions for future research to facilitate the generation of robust, tamper-resistant, and smart automotive black boxes that will be able to assist safety and forensic applications of next-generation vehicles.

Development of a Smart Walking Stick for Blind Persons

2026-06-25T04:40:46+00:00

Visual impairment is a global health challenge that significantly restricts the independence and mobility of affected individuals. Traditional white canes, while widely used, are limited in their ability to detect non-ground-level obstacles or provide real-time navigational assistance. This study presents the design, development, and evaluation of a Smart Walking Stick for blind persons, an Arduino-based assistive device that integrates ultrasonic obstacle detection, vibration and audio alerts, GPS location tracking, and GSM-based emergency communication. The prototype was developed using an Arduino Nano/UNO microcontroller, HC-SR04 ultrasonic sensor, vibrating motor, active buzzer, and a 9 V rechargeable battery unit. Testing confirmed obstacle detection accuracy of approximately 98.8% at close range (5 cm), with detection rates declining at distances beyond 3 m. User trials demonstrated that trained users achieved navigation speeds of up to 0.8 m/s compared to 0.41 m/s for untrained users. The total prototype cost was estimated between ₹1,350 and ₹2,400 (approx. USD 16–29), making the system highly affordable and scalable. Future enhancements include AI-based object recognition, multilingual voice assistance, and IoT integration for remote caregiver monitoring.

LoRaWAN-Enabled AI-Based Smart Surveillance System for Critical Infrastructure Protection

2026-07-31T07:01:54+00:00

This article presents a LoRaWAN-enabled AI-based smart surveillance system designed to enhance the protection of critical infrastructure in remote and resource-constrained environments. Conventional surveillance systems often rely on continuous video streaming, which leads to high bandwidth consumption, large storage requirements, significant power usage, and dependence on reliable communication networks. These limitations make them unsuitable for geographically dispersed sites such as power substations, industrial plants, transportation facilities, and border areas. To address these challenges, the proposed system adopts an event-driven architecture that combines low-power IoT sensor nodes, LoRaWAN communication, edge artificial intelligence, and cloud-based monitoring. The sensor nodes continuously monitor environmental and security-related parameters such as motion, vibration, smoke, gas leakage, temperature, and unauthorized entry. When an abnormal event is detected, the information is transmitted through the LoRaWAN gateway to the edge AI controller, which classifies the event and activates the surveillance camera only when necessary. The captured video is then processed using AI models for object detection and threat verification, and alerts are sent to authorize personnel in real-time. This approach minimizes unnecessary video recording, reduces communication overhead, lowers storage and energy requirements, and improves response time. The proposed framework offers a scalable, cost-effective, and energy-efficient solution for intelligent surveillance in critical infrastructure protection.

Adaptive Quantum-Inspired Optimization for Nonlinear Distortion Compensation in Optical Fiber Networks

2026-07-31T11:24:44+00:00

Nonlinear impairments remain a major challenge in Wavelength Division Multiplexing (WDM) optical fiber networks, particularly at high launch powers and long transmission distances. Conventional compensation techniques such as Digital Backpropagation (DBP) and Machine Learning (ML)-based methods improve transmission quality but often suffer from high computational complexity, extensive training requirements, and limited adaptability. This study proposes an Adaptive Quantum-Inspired Optimization (QIO) framework for mitigating nonlinear impairments in WDM optical fiber communication systems. The proposed approach integrates a quantum-inspired annealing optimization engine with a Nonlinear Schrödinger Equation (NLSE)-based optical transmission model to adaptively optimize launch power, amplifier gain, and digital signal processing parameters. An 8-channel WDM system was modeled in MATLAB and OptiSystem, and its performance was evaluated using the Q-factor, Bit Error Rate (BER), and Optical Signal-To-Noise Ratio (OSNR). The proposed QIO method achieved a peak Q-factor of , compared with for ML, for DBP, and for the uncompensated system. It also produced the lowest BER of outperforming ML ( ), DBP ( ), and the uncompensated system ( at launch power. For long-haul transmission over 500km, the proposed method maintained an OSNR of 19.8 dB, exceeding ML ( ), DBP ( ), and the uncompensated system ( ). These results demonstrate that the proposed QIO framework provides effective nonlinear impairment mitigation while maintaining moderate computational complexity, making it a promising solution for next-generation high-capacity optical fiber communication systems.