secure channel establishment

This work addresses sender privacy in high-bandwidth, scalable anonymous broadcasting by proposing an efficient scheme based on a two-server DC-net architecture. The design incorporates a self-contained anonymous registration protocol that eliminates the need for external dial-up services, a batch messaging mechanism built upon distributed point functions, and a lightweight access control method grounded in secret-sharing-based proofs. By integrating cryptographic primitives such as verifiable distributed point functions, the system achieves millisecond-level registration auditing and supports scaling to thousands of channels. Experimental results demonstrate that the scheme improves effective message throughput by 1.2× to 20× compared to the current state-of-the-art solutions.

Existing agent communication frameworks struggle to simultaneously ensure auditability and privacy, hindering deployment in regulated scenarios such as compliant billing and regulatory accountability. This paper introduces ZK-Audit, a zero-knowledge auditing framework for the Agent Internet, enabling verifiable bidirectional privacy-preserving audits between agents—i.e., verifying compliance with predefined rules and message formats without revealing message content or semantics. Key contributions include: (1) zk-MCP, a zero-knowledge protocol natively compatible with the Model Context Protocol (MCP), guaranteeing authenticity, privacy, and sub-second verification latency; (2) a lightweight zero-knowledge proof system built on Circom; and (3) an asynchronous audit protocol with full implementation. Formal security analysis proves the framework achieves data authenticity and communication privacy; audit latency is negligible. ZK-Audit provides the first practical solution for regulatory compliance, precise usage-based billing, and trustworthy multi-agent collaboration.

This work addresses the weak privacy guarantees of existing wireless service discovery protocols, which often leak sensitive device information. The authors propose PriSrv, a novel protocol built upon a new cryptographic primitive—Anonymous Credential Matching Encryption (ACME)—and a constant-size Fast Anonymous Credential (FAC) scheme. PriSrv employs a two-layer matching architecture: an outer layer filters services using public attributes, while an inner layer enables mutual, fine-grained authentication through selective disclosure of private attributes. This design achieves policy enforcement, selective disclosure, and multi-round unlinkability in a single round and integrates seamlessly with mainstream frameworks such as EAP, mDNS, BLE, and AirDrop. Experimental results demonstrate that PriSrv enables sub-second secure service discovery across multiple platforms, offering strong privacy protection and high usability, with its security formally verified and comprehensively evaluated.

Anonymous communication networks face core challenges including AI-driven metadata tracking, difficulties in decentralized implementation, and the absence of provably secure designs. This paper introduces the first decentralized anonymous routing protocol with a rigorous formal proof of user identity indistinguishability under an extreme adversarial model—eliminating reliance on threshold assumptions or trusted third parties. The protocol integrates cryptographic privacy primitives (e.g., cover traffic routing and differential-privacy-based traffic shaping), distributed topology control, and a formal security verification framework, achieving robust resistance against both passive traffic analysis and active tracing attacks. We formally prove that it satisfies strong anonymity guarantees—namely, k-anonymity and unlinkability. Simulation results demonstrate high throughput and low bandwidth overhead while maintaining sub-millisecond end-to-end latency. Our work establishes a new paradigm for practical, strongly private communication that simultaneously ensures theoretical soundness and deployment feasibility.

To address sender/receiver identity leakage, eavesdropping by intermediaries, lack of key forward secrecy, and unverifiable final states in privacy-preserving communication on the Internet Computer Protocol (ICP), this paper proposes ICPP—the first end-to-end privacy-preserving transport architecture tailored for ICP. Methodologically, it employs dual short-lived intermediaries to decouple access and storage, integrates sealed storage with TEE-based trusted destruction; introduces the first non-interactive encapsulation scheme based on rank-deficient matrix power functions (RDMPF); enables fingerprint-free receiver discovery via publicly known prompts; and ensures auditable final states and liveness verifiability through certified destruction intent and zero-knowledge proofs. Formal verification confirms correctness of encapsulation, prompt privacy, authorization reliability, and timeout-based key revocation. The system is deployed on the ICP mainnet, enabling high-privacy, verifiable, and censorship-resistant secure transport.

Synthetic network traffic generation faces a fundamental trade-off between fidelity and privacy, yet privacy risks of generative models remain poorly quantified. Method: We propose the first comprehensive privacy evaluation framework for network traffic, integrating membership inference attacks, data extraction attacks, and network-specific identifier analysis—including IP flow fingerprints and TLS fingerprints—to quantify sensitive information leakage across diverse generative models and multi-source datasets. Contribution/Results: Experiments reveal up to 88% success rates in membership inference attacks, with certain models fully reconstructing original network identifiers (100% recovery). Model architecture and training data diversity are identified as key determinants of privacy risk. Our findings expose the latent privacy cost of high-fidelity synthetic traffic and establish a verifiable, standardized benchmark for privacy-preserving traffic generation—providing empirically grounded design principles for developing privacy-enhanced synthetic data generators.

While existing agent-to-agent (A2A) interoperability protocols encrypt message content, they inadvertently leak communication graph metadata, enabling adversaries to infer and compromise the integrity of incomplete workflows. This work elevates the threat model of metadata exposure from a privacy concern to a critical workflow integrity risk. We introduce revelatory characteristics centered on semantics, foresight, and execution coupling, and formally define metadata privacy properties at both the transport and orchestration layers. Building upon technologies such as SimpleX/SMP, Tor, and mix networks, we design a metadata-protecting transport binding compatible with A2A protocols. Experimental results demonstrate that raw metadata alone enables high-accuracy task-type classification; however, when our proposed privacy mechanisms are applied, classification performance degrades to random guessing, substantially diminishing adversarial advantage.

Existing covert communication techniques suffer from poor practicality, weak protocol adaptability, and vulnerability to detection. To address these issues, this paper proposes a decentralized, modular covert communication framework. The framework embeds covert data into protocol redundancy fields, integrating traffic camouflage, packet fragmentation, and cross-layer distributed scheduling—enabling multi-protocol adaptive steganography without generating additional packets. Its plug-and-play architecture supports runtime dynamic composition of steganographic channels, significantly enhancing detection resistance. Experimental evaluation on a real-world ten-node network achieves a covert throughput of 1.6 MB/s, demonstrating strong robustness and effective evasion of mainstream intrusion detection/prevention systems (IDS/IPS) and deep packet inspection (DPI) mechanisms.

In federated authentication systems, a privacy-security paradox arises: identity providers (IdPs) must detect cross-domain duplicate or fraudulent registrations while preventing leakage of users’ sensitive attributes or enabling cross-domain linkage. To resolve this, we propose a pseudonymous identifier mechanism based on oblivious pseudorandom functions (OPRFs) and domain-specific transformations. This mechanism enables globally consistent identity verification under full input confidentiality and, integrated with a blinded registration authority, supports a decentralized authentication framework. Experiments demonstrate efficient cross-domain uniqueness verification without exposing raw identity data, significantly mitigating identity spoofing. The scheme simultaneously achieves strong privacy guarantees—formally satisfying *k*-anonymity and unlinkability—and practical anti-fraud utility. It provides a scalable, standards-compatible, privacy-enhancing solution for multi-domain federated authentication.

This work addresses the vulnerability of traditional privacy mechanisms, which often fail catastrophically upon exposure of system components, leading to sensitive information leakage. To mitigate this, the paper introduces a novel paradigm termed Semantic Non-Assemblability (SNA), wherein architectural design ensures that any set of exposed components below a defined threshold cannot reconstruct a meaningful input, thereby preventing inference of sensitive predicates. The approach innovatively incorporates architectural inertia, enabling privacy guarantees to degrade predictably—rather than collapse abruptly—when components are compromised, and integrates organizational audit constraints to enhance practicality. Leveraging a dual-channel provenance architecture, formal verification via ProVerif, structured protocols, and Birthmark-based attestation on constrained hardware, the system achieves unlinkability across devices, observer unidentifiability, server blindness, and correctness of active defense gates, delivering strong, deployable privacy assurances even on resource-limited platforms.