<![CDATA[In a cloud-only Internet of Medical Things (IoMT) system, every urgent health alert must cross the wide area network twice before a clinician sees it. Those two crossings are costly. Under moderate load they use up most of the clinical time budget, and under heavy load most alerts arrive too late to be useful. A fog communication framework is presented here that addresses the problem by moving the alert decision to a gateway near the patient and by giving alert traffic strict priority over routine telemetry. The priority scheme is not a new algorithm. It is the standard three band PfifoFast queue of the Linux traffic control layer, so it can be deployed today on ordinary hardware. A single analytical model joins a log-distance wireless link, a Cobham non-preemptive priority queue at the gateway, and a hard alert deadline, and is solved in closed form for end to end latency, deadline-meet ratio, and backhaul saving. Three independent evaluators are then used: the analytical model, a 25-seed discrete event simulation, and a packet-level NS-3 simulation of the IEEE 802.15.4 access tier running the same PfifoFast queue. Traffic comes from a real vital signs dataset of 200 020 records, whose measured risk balance sets the critical and routine traffic mix. At an offered load of 0.90 the proposed design cuts mean critical-alert latency by 84.0%, from 151.7 ms to 24.2 ms, and raises deadline compliance from 60.6% to 100% against a cloud-only FIFO baseline. A paired Wilcoxon test confirms the difference (p = 5.96 x 10^-8). A factorial ablation shows that fog placement and priority scheduling help for different reasons, and that both are needed. Compared against Earliest Deadline First, which is optimal in theory, the simple stock queue reaches the same deadline reliability at far lower implementation cost. A transient test in which the critical share of traffic rises sixfold shows the guarantee still holds. Backhaul traffic falls by 95.3% at the same time.]]>
Copyrights © 2026
