English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

The Large Language Models (LLMs) are poised to offer efficient andintelligent services for future mobile communication networks, owing to theirexceptional capabilities in language comprehension and generation. However, theextremely high data and computational resource requirements for the performanceof LLMs compel developers to resort to outsourcing training or utilizingthird-party data and computing resources. These strategies may expose the modelwithin the network to maliciously manipulated training data and processing,providing an opportunity for attackers to embed a hidden backdoor into themodel, termed a backdoor attack. Backdoor attack in LLMs refers to embedding ahidden backdoor in LLMs that causes the model to perform normally on benignsamples but exhibit degraded performance on poisoned ones. This issue isparticularly concerning within communication networks where reliability andsecurity are paramount. Despite the extensive research on backdoor attacks,there remains a lack of in-depth exploration specifically within the context ofLLMs employed in communication networks, and a systematic review of suchattacks is currently absent. In this survey, we systematically propose ataxonomy of backdoor attacks in LLMs as used in communication networks,dividing them into four major categories: input-triggered, prompt-triggered,instruction-triggered, and demonstration-triggered attacks. Furthermore, weconduct a comprehensive analysis of the benchmark datasets. Finally, weidentify potential problems and open challenges, offering valuable insightsinto future research directions for enhancing the security and integrity ofLLMs in communication networks.

A Comprehensive Overview of Backdoor Attacks in Large Language Models within Communication Networks