The Bitcoin network is susceptible to various attacks due to its openness, decentralization, and plaintext connections. Bitcoin created a misbehavior-score mechanism for monitoring and tracking peer misconduct. In this paper, we uncover several vulnerabilities of this mechanism, leading to potential Bitcoin-Message-based Denial-of-Service (BitMsg-DoS) attacks on Bitcoin nodes and Slander attacks by maligning innocent nodes. We prototype these attacks for our experiments by testing real nodes connected to the Bitcoin main network (while we do not exfiltrate our attacks to the real-world main network). The experimental results show that the attacks exert varying degrees of impact on mining and non-mining nodes, notably reducing mining rates by up to half for affected mining nodes and decreasing the synchronization speed of blocks for non-mining nodes. To address these drawbacks, this study proposes corresponding countermeasures targeting the identified vulnerabilities in the misbehavior-score mechanism. Furthermore, we explore the Peer-to-Peer (P2P) encrypted transport protocol with experimental support in the latest Bitcoin Core 26.0, but find it insufficient in mitigating the Slander attacks.

Most trust models are identity based, which how-ever are not appropriate to the permissionless peer-to-peer (P2P) networking since anonymity is a built-in property in the cryptocurrency system (e.g., Bitcoin). Hence, there exists an inherent trade-off between anonymity and trust in the context of permissionless P2P networking system. This paper is motivated to propose a keyless authentication based on zero-knowledge proof. With this, peers can authenticate each other without disclosing any sensitive information. To this end, this approach leverages the software-defined networking (SDN) technique to facilitate the zero-knowledge proof so that the peer's link information can be identified while the proving process will never reveal any identity information. Therefore, the challenge-response exchange can prevent Man-in-the-Middle (MITM) attacks with minimal communication overhead. The experimental results built on the prototype show that this approach is efficient.

The MAVLink protocol serves as the cornerstone for control communications between ground control systems (GCS) and unmanned aerial vehicles (UAVs), facilitating essential control communication. Despite its widespread adoption, the protocol's security mechanisms have raised significant concerns. This work focuses on the design vulnerabilities of the MAVLink protocol v2.0 including the deficiency in message authentication code (MAC) mechanism and lapses in sequence number verification. Also, this research reveals the implementation loopholes (as findings) in the well-known GCS software (Mission Planner) for updating the secret key and in the widely used UAV emulation (ArduPilot) for examining invalid timestamps. This breach paves the way for the injection of malicious messages, culminating in the potential hijacking of the UAV. In response to these issues, we propose several countermeasures including a solution using the public key-based signature. The efficacy of both the attack methods and the countermeasures is validated through a series of experiments conducted within a controlled testbed environment.

Capture-the-Flag (CTF) competitions are a prominent method in cybersecurity for practical attack and defense exercises. Despite the rapid advancements in Large Language Models (LLMs), their potential for solving CTF challenges remains underexplored. In this paper, we first propose a flexible CTF platform designed to reflect real-world penetration testing scenarios, bridging the gap between theoretical learning and practical cybersecurity challenges. Our platform is highly customizable, freely deployable, and capable of generating scenarios that closely mirror real network vulnerabilities. More importantly, we introduce an automated LLM agent framework that tackles CTF challenges using an integrated toolchain and various plugins to enhance problem-solving efficiency. In addition, we propose a human-validated LLM agent framework to address the potential limitations of the fully automated LLM agent, providing a clearer evaluation of the LLMs’ intrinsic capabilities. We evaluate the agent’s performance using four LLMs: GPT-4o, GPT-4o mini, o1-preview, and o1-mini. Although the LLMs’ performance on dynamic and complex penetration tests reveals certain limitations, which need further exploration, our experimental results demonstrate that LLMs can leverage their extensive knowledge bases to effectively solve CTF challenges.

Throughout the life cycle of silicon-based system deployment, there may be a service failure induced by external or internal effects (e.g., silicon failure). Dependability of a system is a term to define the ability that the system can avoid such service failure beyond an acceptance level. There is a broad spectrum of attributes of dependability, including reliability, maintainability and safety. Silicon telemetry is a key technique to gain insights of silicon and system in field, and it opens a door to a promising solution space for dependability improvement.