زمانبندي چند هدفه در محاسبات مه با استفاده از الگوريتم فراابتكاري

Healthcare monitoring systems based on the Internet of Things (IoT) utilize wearable sensors, medical devices, an‎d environmental sensors to collect medical data. These systems must provide real-time an‎d highly reliable services while facing challenges such as limited computational resources in the fog layer, device heterogeneity, patient mobility, an‎d strict Quality of Service (QoS) requirements. Since IoT devices lack sufficient processing power, tasks must be offloaded to higher layers with greater resources. In this context, task scheduling plays a vital role, as it is classified as an NP-hard problem an‎d is often solved using metaheuristic approaches to reduce latency an‎d improve performance. In this research, a task scheduling mechanism is proposed for the IoT–fog–cloud environment that simultaneously addresses the requirements of low latency, optimal energy consumption, reduced network load, patient mobility support, an‎d task prioritization. The proposed algorithm is implemented on a four-layer IoT, sink, fog, an‎d cloud framework capable of supporting patient mobility. Medical data such as body temperature, blood oxygen, blood sugar, heart rate, an‎d blood pressure are collected by medical devices. First, the urgency level of each task is calculated, an‎d tasks are classified into two queues: “critical” an‎d “non-critical.” Each queue is then ordered using the Weighted Sum Model (WSM), based on maximum response time an‎d urgency level. For scheduling, a Meta-Heuristic Adaptive Hybrid Scheduler for Healthcare (MHASH) is introduced, which modifies the boundary check mechanism an‎d employs a weighted fitness function with different weights for latency an‎d energy in critical an‎d non-critical tasks, as well as penalties for invalid allocations. To avoid local optima, Harris Hawks Optimization (HHO) is hybridized with Simulated Annealing (SA) using an intelligent neighbor selec‎tion mechanism. Critical tasks are prioritized for allocation to fog nodes, while non-critical tasks are distributed opportunistically between fog an‎d cloud nodes depending on available capacity. The proposed algorithm is simulated in the iFogSim environment under two conditions: without queuing delay an‎d with queuing delay. The SA temperature function is implemented in three variants: exponential, linear, an‎d logarithmic. Each case is compared with Mobility-aware Modified Balance an‎d Reduction (MobMBAR), Priority-based Task Scheduling an‎d Resource Allocation (PTS-RA), an‎d baseline HHO algorithms. Results demonstrate that overall latency is reduced by up to 10.2% without queuing delay an‎d up to 13.6% with queuing delay. Energy consumption decreases by up to 73.9% without queuing delay an‎d up to 47.1% with queuing delay, while network load is reduced by up to 57.8% without queuing delay an‎d up to 63.7% with queuing delay.

6

149768