Electronics · 20 Nov 2026
Resource-Adaptive Lightweight Security-Agent-Based Collaborative Intrusion Detection for Resource-Constrained Tactical Terminals
Xiaoyue Bai, Hao Shi, Lei Li, Fengwei Sun
Citations by database
Each database indexes a different set of journals, preprints and books, so they are shown side by side rather than blended.
- OpenCitations0
- Crossref0
Citations per year
The current year is still counting.
- 2026 so far0
Abstract
Resource-constrained tactical Internet of Things terminals must run intrusion detection under changing computation, memory, and latency budgets, and each observes only part of a distributed incident. This paper proposes Lightweight Resource-Adaptive Collaborative Detection (LRACD), a terminal-level security agent whose core is a two-stage ordered mechanism: feasibility tested per resource dimension, collaboration triggered by predictive uncertainty, receiver-estimated reliability, and a bound on the aggregate neighbor influence. On held-out LSPR24 windows the Full exit reaches a macro-F1 of 0.922 with 5768 active parameters, against 0.926 for a Transformer carrying 8.1 times as many; under time-varying budgets LRACD holds decision coverage of 99.5% at a 0.5% deadline-violation rate and 10.9 ms p95 decision latency, where the strongest fixed-mode policy falls to 88.3% coverage at 11.7% violation and 24.8 ms; and in collaborative episodes it reaches an event recall of 0.907 at 68 application bytes per node-second, 63.2% below continuous collaboration. The separation that produces these numbers is a division of labor: the computation mode follows the resource state alone, and communication is triggered by the predictive uncertainty and admitted subject to the communication budget. A shared temporal network exposes a shallow Lite exit and a complete Full exit, so the mode is binary; a feasibility filter tests each dimension against its own budget and a hysteretic scheduler ranks only the surviving modes, so no binding constraint is averaged away and no prediction is needed to choose the exit. The calibrated uncertainty then decides whether neighbors are consulted, and the receiver fuses admitted, in event-aligned reports through a bounded convex combination weighted by its own reliability record and the declared uncertainty, with the aggregate neighbor contribution capped. Because the exit is fixed before any prediction exists, no prediction can feed back into the choice that produced it.
What the data shows
- No database records a citation yet, as is usual for a paper this new.
Graph
This paper in the middle, its authors and their most-cited papers below, the papers it cites to the left and the papers citing it to the right. The timeline lays every paper out by year.
Select a node for its details. Drag to pan, scroll to zoom, drag a node to move it.
Start with
Cited by 0 papers
Who cites this paper, from open citation data, with Crossref's record of each; the 100 newest when there are more.
References (20)
What it cites, from open citation data, with Crossref's record of each.