Sponge Attacks on Sensing AI: Energy-Latency Vulnerabilities and Defense via Model Pruning

TL;DR

This paper investigates energy-latency sponge attacks on sensing AI models in IoT devices, demonstrating their impact and proposing model pruning as an effective defense to enhance resilience while analyzing efficiency-security trade-offs.

Contribution

It is the first systematic study of sponge attacks on sensing AI in resource-constrained environments and explores model pruning as a defense mechanism.

Findings

Sponge attacks significantly increase energy consumption and latency in sensing AI.

Model pruning improves resilience against sponge poisoning attacks.

Trade-offs exist between model efficiency and attack resilience.

Abstract

Recent studies have shown that sponge attacks can significantly increase the energy consumption and inference latency of deep neural networks (DNNs). However, prior work has focused primarily on computer vision and natural language processing tasks, overlooking the growing use of lightweight AI models in sensing-based applications on resource-constrained devices, such as those in Internet of Things (IoT) environments. These attacks pose serious threats of energy depletion and latency degradation in systems where limited battery capacity and real-time responsiveness are critical for reliable operation. This paper makes two key contributions. First, we present the first systematic exploration of energy-latency sponge attacks targeting sensing-based AI models. Using wearable sensing-based AI as a case study, we demonstrate that sponge attacks can substantially degrade performance by increasing energy consumption, leading to faster battery drain, and by prolonging inference latency. Second, to mitigate such attacks, we investigate model pruning, a widely adopted compression technique for resource-constrained AI, as a potential defense. Our experiments show that pruning-induced sparsity significantly improves model resilience against sponge poisoning. We also quantify the trade-offs between model efficiency and attack resilience, offering insights into the security implications of model compression in sensing-based AI systems deployed in IoT environments.