Deep Learning Latency Attacks and Defenses: A Cross-Domain Survey of Availability Threats

2026-09-29

AI Summary

This work addresses availability threats to deep learning systems, such as response timeouts and resource exhaustion induced by latency attacks. Conducting a cross-domain survey spanning perception, inference, and agent systems, it transcends conventional attack-form taxonomies by unifying fragmented literature through the lens of computational bottlenecks. The study proposes a defense abstraction based on work budgets alongside a system-level failure analysis framework, integrating techniques from adversarial machine learning, adaptive neural inference, and large model security for in-depth examination. Ultimately, it establishes a threat model taxonomy, provides quantitative comparisons and a comprehensive review of defense strategies, and delineates open challenges for future research.

Abstract

Adversarial machine learning has focused mainly on integrity, but availability is an increasingly consequential complement. Latency attacks (also energy-latency attacks) increase inference-time work, energy, or response time, causing deadline misses, throughput collapse, or resource exhaustion in vehicle controllers, interactive services, or battery-powered sensors, sometimes while preserving the nominal prediction. This survey unifies a fragmented literature spanning perception pipelines (including physical attacks on autonomous-driving detection and tracking), input-adaptive neural inference (sponge examples, dynamic networks), and autoregressive and agentic systems (output-length, verbose-image, and reasoning denial-of-service attacks on LLMs, VLMs, mixture-of-experts models, and tool-using agents). We organize attacks by exploited computational bottleneck rather than formulation, separating what makes a computation expensive from how the attacker triggers it; the delivery channel (input, prompt or retrieved content, message, poisoning, or weight tampering) is an orthogonal attribute. Many attacks share one mechanism, intermediate-work amplification, motivating a work-budget defense abstraction; we distinguish caps on the work entering an expensive stage from caps on the results leaving it. We further analyze when a model-level cost increase becomes a system-level availability failure, which depends on critical-path share, slack, existing ceilings, accumulation, resource sharing, and fallback policy, not on the amplification factor alone. We also provide a threat-model taxonomy, consolidated quantitative comparisons, a defense review by control mechanism, and open challenges such as standardized evaluation, physical realizability, and whole-system availability. Companion website: https://github.com/guzonghua/awesome-latency-attacks.

Problem

• Latency attacks
• Deep learning availability
• Adversarial machine learning
• Denial-of-service
• Inference-time resource exhaustion

Innovation

• Latency Attacks
• Intermediate-Work Amplification
• Work-Budget Defense
• System-Level Availability
• Computational Bottleneck