LiFeChain: Lightweight Blockchain for Secure and Efficient Federated Lifelong Learning in IoT

Abstract

Internet of Things (IoT) devices constantly generate heterogeneous data streams, driving demand for continuous, decentralized intelligence. Federated Lifelong Learning (FLL) provides an ideal solution by incorporating federated learning and lifelong learning. However, the extended lifecycle of FLL in IoT systems increases their vulnerability to persistent attacks. This problem is exacerbated by the single point of failure. Furthermore, the single point of trust created by the central server hinders reliable auditing for long-term threats. Blockchain technology provides a tamper-proof foundation for trustworthy FLL. Nevertheless, directly applying blockchain to FLL significantly increases computational and retrieval costs with the expansion of the knowledge base, slowing down the training on resource-constrained IoT devices. To address these challenges, we propose LiFeChain, a lightweight blockchain for secure and efficient federated lifelong learning with minimal on-chain disclosure and bidirectional verification. LiFeChain is the first blockchain tailored for FLL. It incorporates two complementary mechanisms: the Proof-of-Model-Correlation (PoMC) consensus on the server, which couples learning and unlearning mechanisms to mitigate negative transfer; and Segmented Zero-knowledge Arbitration (Seg-ZA) at the client, which detects and arbitrates abnormal committee behavior without compromising privacy. LiFeChain is a plug-and-play component that can be seamlessly integrated into existing FLL algorithms for IoT applications. To demonstrate its practicality and performance, we implement LiFeChain in representative FLL algorithms with Hyperledger Fabric under 6 attacks. Theoretical analysis and extensive evaluations demonstrate that LiFeChain effectively mitigates long-term attacks, and significantly reduces latency and storage overhead compared to state-of-the-art blockchain solutions.

Figures (28)

• Figure 1: Comparison of FL, LL, and FLL. FL shares knowledge from multiple clients for one task, resulting in spatial heterogeneity across clients. LL accumulates knowledge from past tasks within a single client, leading to temporal heterogeneity over time. FLL integrates knowledge across both clients and tasks, imposing substantial storage demands for managing a continuously expanding knowledge base, and introduces spatial-temporal heterogeneity.
• Figure 2: The architecture of LiFeChain consists of three primary components: clients, a committee, and a blockchain. C and S represent client- and server-side steps, respectively. The steps in each training round are detailed as follows: [C1] Client receives the aggregated model; [C2] Client queries LiFeChain to retrieve knowledge using KRV; [C3] Client fuses the aggregated model with the retrieved knowledge for local training; [S4] Server selects and aggregates the global model; [S5] Server computes the KRVs of knowledge; [S6] Server uploads the generated blocks to LiFeChain if the blocks are validated through PoMC; [C7] (Optional) Client initiates an arbitration to validate committee behavior.
• Figure 3: The workflow of LiFeChain.
• Figure 4: The workflows of Seg-ZA.
• Figure 5: Latency for a task in networks of different sizes. The network sizes are represented as: num_clients(num_committee_servers). (a) FedKNOW. (b) FedWeIT.

Theorems & Definitions (23)

• Definition 1: Memory Contamination Attacks
• Definition 2: Client-side Adversary $\mathcal{A}_{client}$
• Definition 3: Server-side Adversary $\mathcal{A}_{server}$
• Theorem 1: Optimization of Descent under Gradient Conflict
• Theorem 2: Forgetting Minimization on Dominant Tasks
• Lemma 1: Quantization Fidelity and Tolerance Bound
• Theorem 3: Completeness of Seg-ZA under Error Bound
• Theorem 4: Computational Soundness against Malicious Aggregation
• Theorem 5
• Theorem 6