Scholarlink

<p>Edge computing brings cloud capabilities closer to users via heterogeneous IoT devices, but this also enlarges the attack surface and complicates security and compliance. This paper reviews and organises threats against edge-based IoT infrastructures and surveys mechanisms for secure data transmission, device authentication, trusted execution and monitoring. Building on these insights, we propose the EOTIS (Edge Offloading Threat Intelligence and Security) framework to integrate secure offloading, federated intrusion detection and compliance-aware orchestration for resilient edge deployments. We further instantiate the proposed EOTIS framework in a federated intrusion detection prototype over synthetic edge-IoT traffic (100k flows, 10 edge nodes), comparing centralised, local-only and federated models. In this prototype, EOTIS-FL improves attack recall by about 6 percentage points (roughly 12% relative) and F<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(_1\)</EquationSource> </InlineEquation>-score by about 4 percentage points (around 6% relative) over a centralised logistic regression baseline, while reducing communication overhead by more than 99% and significantly increasing detection rates for DDoS and exfiltration attacks.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Resilient Edge Computing Infrastructures with Secure Data Offloading for IoT Workloads

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Vikas Shukla,
  • Rekha Agarwal,
  • Rajesh Kumar Tyagi

摘要

Edge computing brings cloud capabilities closer to users via heterogeneous IoT devices, but this also enlarges the attack surface and complicates security and compliance. This paper reviews and organises threats against edge-based IoT infrastructures and surveys mechanisms for secure data transmission, device authentication, trusted execution and monitoring. Building on these insights, we propose the EOTIS (Edge Offloading Threat Intelligence and Security) framework to integrate secure offloading, federated intrusion detection and compliance-aware orchestration for resilient edge deployments. We further instantiate the proposed EOTIS framework in a federated intrusion detection prototype over synthetic edge-IoT traffic (100k flows, 10 edge nodes), comparing centralised, local-only and federated models. In this prototype, EOTIS-FL improves attack recall by about 6 percentage points (roughly 12% relative) and F \(_1\) -score by about 4 percentage points (around 6% relative) over a centralised logistic regression baseline, while reducing communication overhead by more than 99% and significantly increasing detection rates for DDoS and exfiltration attacks.