Blockchain systems, such as Bitcoin and Ethereum 2.0, face vulnerabilities under bandwidth-constrained partitions, where throughput collapses and latency increases. In addition, adversaries can exploit inconsistencies to launch double-spending attacks. This study presents a lightweight dual-layer countermeasure that integrates a robust freezing threshold ( ) with multi-signal disconnection proofs to enhance performance and security without altering consensus rules. Controlled simulation experiments on Bitcoin (PoW) and Ethereum 2.0 (PoS) show throughput gains exceeding 1000% in Ethereum and over 100% in Bitcoin, with inconsistency reduced by up to 64% and latency bounded within 5-6 blocks/s. These results confirm that attacker-aware thresholds and multi-signal validation substantially improve blockchain resilience under partitioned network conditions.
Reaching consensus in Proof-of-Stake (PoS) based consensus protocols, requires supermajority agreement among participating validator nodes. Such protocols need significant network resources due to the concurrent voting of a large number of consensus nodes. As a solution, these nodes are divided into committees, with each committee voting individually at a dedicated time slot. In this paper, we introduce CliqueSensus, a protocol that, given a distribution of consensus nodes into committees, lets them self-organize into small, ephemeral clusters structured in clique topologies, to accelerate the voting process, while using only a small fraction of the network resources required by conventional message dissemination methods. Our evaluation demonstrates that our protocol exhibits rapid convergence and operates with minimal network overhead. We focus on the PoS consensus algorithm adopted by Ethereum 2.0. In addition to our protocol, we also analyze and simulate the clustering approach that Ethereum has adopted, showcasing that our protocol can reduce validation message dissemination time by 23% to 70%, while requiring about 190 times fewer validation message forwards.
The Proof of Stake (PoS) consensus mechanism is increasingly used in blockchain systems; however, resource allocation for PoS-based mobile blockchain networks remains underexplored, particularly given the constraints of mobile devices. This work introduces MEC-Chain, a new framework that integrates Mobile Edge Computing (MEC) with mobile blockchain to support efficient validator-node execution under PoS. MEC-Chain formalizes a multi-objective resource-allocation problem that jointly considers latency, reliability, and cost from both the validator and MEC-provider perspectives. To address this challenge, we develop a deep reinforcement learning-based allocation agent using the Proximal Policy Optimization (PPO) algorithm. Experimental results show that PPO achieves a 30–40% reduction in total execution time, 25–35% lower transmission latency, and 10–15% higher reliability compared to A2C (Advantage Actor–Critic) and DQN (Deep Q-Network), while offering comparable cost savings across all methods. These results demonstrate the effectiveness of MEC-Chain in enabling low-latency, reliable, and resource-efficient PoS validation within mobile blockchain environments.
Blockchain networks face a critical but understudied threat: wave attacks that exploit difficulty adjustment algorithms through strategic mining participation. Adversaries cyclically withdraw and re-enter mining to create oscillations that degrade network liveness and destabilize honest miners’ revenue. We present the first production-ready framework that maintains network responsiveness while enabling robust, post hoc threat detection. The framework employs a statistically rigorous pipeline featuring controller-aligned anomaly detection, transitive collusion grouping via union-find, and Benjamini–Hochberg False Discovery Rate control. We formally prove the economic viability of this architecture: when penalties on unvested rewards are enabled by governance, wave attacks become asymptotically unprofitable for rational adversaries. Evaluated on a 128-node distributed testbed simulating Bitcoin, Ethereum Classic, and Monacoin networks over 30 independent runs, our framework achieves 92.7% F1-score in detecting attacks, significantly outperforming baseline methods (74.7%). This work provides a complete, theoretically-grounded solution for securing proof-of-work blockchains against difficulty manipulation, forming the foundation for the adaptive AI-driven enhancements presented in our companion paper (Part II).
Decentralized multi-agent systems have shown promise in enabling autonomous collaboration among LLM-based agents. While AgentNet demonstrated the feasibility of fully decentralized coordination through dynamic DAG topologies, several limitations remain: scalability challenges with large agent populations, communication overhead, lack of privacy guarantees, and suboptimal resource allocation. We propose AgentNet++, a hierarchical decentralized framework that extends AgentNet with multilevel agent organization, privacy-preserving knowledge sharing via differential privacy and secure aggregation, adaptive resource management, and theoretical convergence guarantees. Our approach introduces cluster-based hierarchies where agents self-organize into specialized groups, enabling efficient task routing and knowledge distillation while maintaining full decentralization. We provide formal analysis of convergence properties and privacy bounds, and demonstrate through extensive experiments on complex multi-agent tasks that AgentNet++ achieves 23% higher task completion rates, 40% reduction in communication overhead, and maintains strong privacy guarantees compared to AgentNet and other baselines. Our framework scales effectively to 1000+ agents while preserving the emergent intelligence properties of the original AgentNet.
Jorge Soria, Jorge Moya Velasco, Carlos Estévez-Mendoza
Decentralized Physical Infrastructure Networks (DePIN) represent an emerging organizational form for operating physical infrastructure through blockchain-based coordination. DePIN through decentralized protocols and token-based payment mechanisms incentivize independent agents to deploy, maintain, and monetize real-world infrastructure, such as wireless networks, storage units, or sensors. This article presents a first formal economic analysis of DePIN architectures, modelling investment decisions under network effects in a blockchain-native Decentralized Autonomous Organization (DAO), with protocol-defined reward schemes. It establishes the equilibrium conditions that support decentralized provision, where token prices internalize participation, service reliability, and network coverage. Furthermore, it identifies a minimum viable coverage threshold determined by costs and network effects. Through a multi-agent machine learning simulation, we confirm that decentralized provision improves efficiency compared to centralized models. The results support the economic viability of DePIN and provide design guidelines for future decentralized infrastructure protocols. Finally, we propose an DAO incentive mechanism to implement First Best provision in Decentralized Physical Infrastructure Networks.
Dynamic spectrum sharing (DSS) is essential for 6G networks, yet existing blockchain-based DSS solutions often lack an integrated approach that simultaneously addresses trust, allocation fairness, and system scalability. This paper proposes HierSpectrumChain, a hierarchical blockchain framework that incorporates a global main chain, localized sub-chains, and a smart-contract based Stackelberg auction for credible and automated spectrum allocation. The system model formalizes interactions among spectrum holders, secondary users, and sub-chain validators, enabling transparent bidding and decentralized coordination. A proof-of-concept implementation on an Ethereum Ganache environment evaluates the functional correctness of the auction workflow and measures throughput under varying client loads. While the evaluation is limited to a single-node testbed, the results demonstrate the feasibility of the proposed architecture and establish a basis for future multi-peer experiments on permissioned blockchains. This work provides a coherent design and initial validation for blockchain-enabled DSS in 6G networks.
The proliferation of Web3 and Internet of Things (IoT) applications generates unprecedented volumes of real-time data streams, demanding secure and efficient subscription mechanisms that uphold data sovereignty. While decentralized architectures are the logical paradigm to ensure this sovereignty, a prominent class of existing schemes suffers from critical vulnerabilities—notably revocation attacks and prohibitive communication overhead—that severely hinder their practical deployment in large-scale environments. This paper introduces SegSub, a novel decentralized data subscription scheme specifically designed to significantly enhance both security and efficiency. SegSub's core innovations include the Segmented Dual-Key Regression with Binary Hash Trees (SDKR-BHT) mechanism, which partitions key regression chains into isolated segments to effectively contain potential data leakage and optimize token management and a strategic user grouping policy that localizes key updates, thereby substantially reducing system-wide communication overhead during revocation events. We formally quantify security improvements using a proposed security index and demonstrate a configurable trade-off between security and efficiency. Theoretical analysis and extensive experimental results validate that SegSub's security index is inversely proportional to segment length while communication efficiency is directly proportional. Furthermore, our grouping policy significantly reduces communication costs in large-scale scenarios through optimal group sizing. SegSub offers a robust and adaptable foundation for sovereignty-preserving data subscription services in Web3, empowering system designers with precise control over the critical security-efficiency balance to meet diverse deployment requirements.
Zubaida Rehman, Mark A. Gregory, Iqbal Gondal, Hai Dong · 5 authors
Ethereum has emerged as one of the most widely used blockchain platforms, underpinning decentralized finance, smart contracts, and distributed applications. With its growing adoption, the Ethereum peer-to-peer network is susceptible to networklayer attacks including eclipse (node-isolation) attacks. To study the threats to Ethereum and to develop effective detection and mitigation strategies, researchers require controlled, reproducible, and labeled network datasets. However, datasets are scarce due to the complexity of capturing live blockchain traffic and the difficulty of confidently labeling malicious activity on public networks. In this paper, we present the design and deployment of a private Ethereum testbed for dataset collection. Our testbed consists of five virtual machines running Geth clients interconnected via a controlled gateway: four nodes act as benign Ethereum peers and one node acts as a malicious entity that performs eclipse attacks. The testbed emulates normal blockchain operations (block propagation, transaction exchanges, and peer discovery) and adversarial scenarios focused on node isolation. Wireshark is deployed on the gateway to capture the network traffic, enabling us to record raw packet traces for benign and attack scenarios. The resulting dataset provides a comprehensive view of Ethereum network-layer behavior, with traffic labeled according to ground truth (node role and attack phase). We describe the testbed, the attack procedure for generating eclipse conditions, the capture and labeling pipeline, and potential uses of the dataset for intrusion detection and resilience analysis.
Sasikumar Asaithambi, Sunil Prajapat, Mohammed Wasim Bhatt, Syed Rizwan Hassan
ABSTRACT The Internet of Things (IoT) can offer more precise, intelligent, and low‐ or non‐human involvement approaches to various sectors. One of the significant uses of the IoT is in smart cities, which includes a variety of services, including smart home, garbage disposal, and smart grid. Many different collaborative IoT integrations are available in smart cities because of these diverse services. A digitized smart city was created to offer complete government cooperation solutions based on digitization and automation to improve residents' quality of life. Information safety and privacy concerns arise when various services need to work together seamlessly. Trustworthy data is vital to the federal government and its constituents, and data accuracy and privacy must be ensured. In this work, we presented a smart contract‐enabled smart city and software‐defined networking (SDN) in limited contexts during collaborative activities based on a controlled network and decentralization. The proposed collaborative application safety structure is being tested on the Hyperledger blockchain networks. We describe a unique approach to data security through collaborative work in intelligent city governmental design, utilizing Proof‐of‐Trust Collaboration (PoTC) in Hyperledger blockchains. A security approach based on SDN and smart contracts is employed to safely manage and monitor all connections and transactions across diverse IoT networks. To assess the viability of the proposed decentralized security framework, we created a supported scenario for collaborative activities in an SDN‐enabled IoT design. We have conducted various experimental simulations to test the proposed blockchain‐integrated SDN‐based IoT architecture for a smart city, including throughput, access delay, and trust evaluation. The simulation results show that the proposed SDN‐enabled blockchain networks provide better results than existing works.
Smart contracts suffer from critical vulnerabilities such as reentrancy attacks, delegate call misuse, and timestamp dependencies, posing significant risks when deployed in smart grids. This paper proposes a solution that integrates deep learning techniques based on the Graph Isomorphism Network (GIN) to analyze Solidity code and detect vulnerabilities before deployment. This model was trained on a dataset of vulnerable contracts, leveraging automated hyperparameter optimization via Optuna to fine-tune the model’s performance. Performance evaluations supported by visual analysis of the model’s pre- and post-training weights provide essential insights into the behavior of the proposed approach. The experimental results demonstrated the effectiveness of GIN in identifying vulnerabilities with an accuracy of 95.26%, outperforming previous studies by more than 2%. These new findings highlight the potential of deep learning to secure and develop resilient smart grid infrastructures.
Verifiable network telemetry is crucial for ensuring transparency and trust in network measurements. However, telemetry logs (e.g., NetFlow records) often contain sensitive data, making public verification challenging. Recent work has attempted to address this problem using Trusted Execution Environments (TEEs), such as Intel SGX, to provide confidentiality and integrity guarantees. However, TEEs are known to suffer from complex deployment requirements and limited scalability. In this paper, we introduce a software-based approach utilizing the latest advances in Zero-knowledge Proofs (ZKPs) to enable verifiable network telemetry without revealing the underlying sensitive logs or relying on special-purpose hardware. Our system employs a general-purpose ZKP virtual machine (RISC Zero) to generate cryptographic proofs over NetFlow data, enabling operators to securely attest to network flow metrics. Our preliminary results indicate that our ZKP-based design offers a viable path toward overcoming deployment and scalability limitations inherent in the solutions that require special-purpose hardware.
Smart contracts are a core component of blockchain ecosystems, but their transparency and immutability make them vulnerable to attacks, leading to significant financial losses. Thus, repairing vulnerabilities in smart contracts is crucial for establishing a trustworthy blockchain environment. Existing smart contract vulnerability repair methods suffer from a critical "one-for-all" design limitation, where a single model is tasked with fixing diverse vulnerability types, leading to suboptimal performance due to insufficient specialization. To address this, we propose MoEFix, a novel framework leveraging a Mixture-of-Experts (MoE) architecture tailored for smart contract characteristics. MoEFix partitions vulnerabilities into subspaces, trains specialized experts for each type (e.g., reentrancy, integer overflow), and employs a vulnerability-aware router to dynamically allocate repairs. We further redesign the repair workflow to align with large language models, enabling end-to-end secure contract generation instead of partial patches, and to achieve this, we curated a dataset of 1,391 contracts covering five critical vulnerability types.To validate our approach, we extend the benchmark PVD test suite. Experiments demonstrate that MoEFix outperforms state-of-the-art methods by 21.64% in overall accuracy, achieving improvements of 26.19% (reentrancy) and 23.08% (delegatecall) for specific vulnerabilities.
Abstract Distributed Software Define Network (DSDN) offers to eliminate the single point of failure present in the centralized SDN controller. The main concept is to have numerous controllers that can share the burden on the network, and one controller can take over another controller when it breaks. In this study, we work on an Opendaylight-based completely distributed system to deal with issues of the controller's central processing unit (CPU) consumption being uneven. Every controller administers its own domain therefore sharing the burden across the network's controllers. To update the flow rules and broadcast new rules to all the controllers at once, maintaining the consistency of the controller's rule set, we create an application ledger that maintains a distributed ledger. Results demonstrate that, in addition to ensuring the uniformity of inventory shards across all controllers in the system, our work presented an increase with throughput reaching 40% and the bandwidth of controller's communication is less with almost 31%.
Collin Arnold Kabwama, Osorachukwu Maurice Ayozie, Justin Njimgou Zeyeum, Adeniran Oluwatoyosi Awe · 6 authors
As Critical National Infrastructure (CNI) becomes increasingly digitized, traditional reactive security assessments are failing to keep pace with automated threats. This paper proposes an autonomous framework integrating Generative Adversarial Networks (GANs) and Distributed Ledger Technology (DLT) for proactive vulnerability discovery and immutable infrastructure hardening. We utilize a Physics-Aware Wasserstein GAN to synthesize protocol-specific attack vectors that identify "zero-day" weaknesses in Industrial Control Systems (ICS) by exploring the operational state space of protocols like DNP3 and Modbus. Discovered vulnerabilities are committed to an Immutable Knowledge Base (IKB) on a sharded, permissioned blockchain, providing a "Single Source of Truth" for cross-sector threat intelligence. To automate mitigation, Smart Contracts orchestrate infrastructure hardening by validating patches through digital twin simulations before network-wide deployment. Experimental results using HELICS and NS-3 demonstrate that this architecture reduces the Mean Time to Remediate (MTTR) from days to sub-second intervals. Finally, we address long-term security by incorporating Post-Quantum Cryptography (PQC) to protect the ledger against emerging quantum threats.
Open access
Smart Grid Security and Resilience
Software-Defined Networks and 5G
Infrastructure Resilience and Vulnerability Analysis
Blockchain technology relies on decentralized data management across various applications to ensure secure transactions. Consensus algorithms play a critical role in maintaining consistency and trust within decentralized environments, laying the groundwork for further exploration. This paper presents a simulation-based benchmarking framework for evaluating major consensus mechanisms Proof of Work (PoW), Proof of Stake (PoS), Proof of Authority (PoA), Proof of Capacity (PoC), Practical Byzantine Fault Tolerance (PBFT), Directed Acyclic Graph (DAG), and Proof of Biometric Uniqueness (PoBU). A MATLAB-based simulation environment integrated with Docker-emulated blockchain nodes was developed to measure performance across several parameters, including throughput, latency, energy consumption, finality probability, fault tolerance, and resource requirements. Results show that DAG achieves the highest throughput (1600 TPS) and lowest latency (0.8 s), while PoBU offers strong fault tolerance (50%) and energy efficiency (60 J/tx). PBFT and PoA provide ultra-fast confirmations with limited decentralization, whereas PoW demonstrate poor scalability and energy efficiency. These findings highlight the potential for a unified benchmarking framework for consensus mechanisms and provide practical insights for selecting consensus protocols for specific blockchain applications.
Jinwen Liang, Jiannong Cao, Bo Yang, Dongbin Bai · 6 authors
Decentralized Physical Infrastructure Networks (DePIN) represent a paradigm shift in infrastructure deployment and resource coordination, leveraging blockchain and token-based incentives to transform traditional service models. This paper presents a comprehensive survey of DePIN, encompassing its evolution, architecture, open issues, and practical implementation. We begin by tracing the development of DePIN across three key phases: the emergence of fungible tokens, the rise of non-fungible tokens (NFTs), and the shift toward real-world asset tokenization. We highlight representative DePIN projects and propose a unified four-layer architecture comprising the infrastructure, decentralized data, resource control, and application layers, each delivering essential functionality for scalable, interoperable DePIN systems. We then analyze open research challenges within each layer and outline promising future research directions. To demonstrate the viability of our architecture, we present DCEAI, a prototype DePIN platform that enables decentralized LLM inference via edge computing. We evaluate its performance and discuss opportunities for further enhancement. This survey aims to establish a solid foundation for future research and innovation in DePIN ecosystems.
E. Orestes O’Brien, Breno da Costa Paulo, Ángel Martín, Muhammad Shuaib Siddiqui · 5 authors
Future networks will deliver unprecedented performance, versatility, and efficiency, driven by advancements in automated monitoring and control. As networks evolve into interconnected platforms capable of enabling advanced features, allocating assets, and configuring them via northbound Application Programming Interfaces (APIs), dynamic control over multiple actors and stakeholders becomes essential. At this point, identity management plays a crucial role in overseeing onboarding, administering access, and registering activity to log and account for all operations. This paper compares the Open Common API Framework (CAPIF), based on the European Telecommunications Standards Institute (ETSI) standard, with a novel distributed identity-based approach using Distributed Ledger Technology (DLT). To assess the viability of these solutions, various tests are conducted to evaluate the frameworks’ latency and processing workloads.
Luigi Mastromauro, Muslum Ozgur Ozmen, Michel A. Kinsy
The increasing complexity of decentralized IoT and edge environments requires systems capable of real-time topological self-organization, autonomous role assignment, and adaptive resilience under dynamic and unpredictable conditions. However, current approaches often rely on static structures, centralized orchestration, or periodic reevaluation, limiting their scalability and robustness. In this work, we propose AtoNet, a fully decentralized and adaptive algorithm for dynamic topology management in IoT networks. AtoNet leverages behavioral validation, trust-based role assignment, and inter-agent coordination to ensure resilient structure formation and secure, autonomous operation. The system includes real-time event detection, fault tolerance via heartbeat-based monitoring, and local topology reconfiguration triggered by trust decay or network stress. Experimental simulations demonstrate that AtoNet maintains low latency, high throughput, and fast adaptation rates, even in highly volatile or congested scenarios, highlighting its potential applicability in decentralized edge-IoT contexts.
Деменко, Євгеній Євгенович, Гребеннік, Ігор Валерійович, Колмиков, Максим Миколайович
The problem of reproducibility of experiments in optimizing validator allocation in blockchain networks with Proof of Stake consensus was investigated, in particular due to the absence of standardized datasets and unified testing methods, which complicates the objective comparison of algorithms. To tackle this issue, we propose a method for building test datasets that rely on deterministic pseudorandom sequence generators and validator profiles calibrated against Ethereum network statistics. Each validator is described by a set of parameters that includes the stake size with the minimum requirement according to Ethereum standards, performance with a uniform distribution, reliability in a high range, network delays depending on the geographical proximity of participants, geographical location according to the actual statistics of validator distribution by regions, quality of network connection, and slashing history according to the violation statistics in the Beacon Chain. Three datasets of different scales were created for small, medium, and large network configurations with fixed initial values of the generators to ensure full reproducibility of experiments. A multi-criteria evaluation system was developed based on a generalized quality indicator that maximizes system throughput and minimizes load imbalance and network delays with scientifically grounded weighting coefficients. The tenfold testing protocol ensures the statistical reliability of results and reduces the impact of randomness on conclusions. The experiments conducted a comparative analysis of four allocation algorithms: a hybrid metaheuristic method based on particle swarm optimization with local search, random allocation with correction, an adapted Ethereum shuffling mechanism, and a greedy algorithm. The experimental results revealed scale-dependent efficiency of the algorithms: the hybrid method provides high optimization quality at all investigated scales, but quadratic growth of execution time limits its application to periodic offline planning of network configuration; the shuffling mechanism demonstrates stable medium-quality results with fast execution; the random method is characterized by moderate speed with variable results; the greedy algorithm shows maximum speed with deterministic results but variable efficiency depending on the network scale. The proposed method forms a basis for standardizing experimental research in Proof of Stake consensus systems. It ensures the objective comparison of new algorithmic solutions for validator allocation in decentralized blockchain networks.
Ilya Evdokimov, Stepan Kamentsev, A. V. Alexeev, Kirill Mikheev · 5 authors
This paper surveys the evolution of smart contract virtual machines (VMs), highlighting major designs such as Ethereum Virtual Machine (EVM), Move VM, Solana VM and couple of lesser known VMs. It eventually proposes a Rust implementation of Ton Open Network Virtual Machine (TVM) called Tycho VM. We analyze the theoretical foundations behind TVM’s “everything is a cell” paradigm and message execution lifecycle inside sandboxed VM. We also contextualize the role of virtual machines in the broader blockchain ecosystem by referencing developer adoption metrics and indicate challenges arising from that, namely, inefficiencies of Solidity to Tycho VM compiler which are going to be addressed in the future research.
G will enable sophisticated composite services that can be realized, bought, and sold across multi-stakeholder marketplaces. With the advent of intent-based networking, requests for these composite services will be through high-level intents, together with requisite service-level agreements (SLAs), and with significant latitude on the specifics of how the composite services are realized. At the same time, the 6 G ecosystem will be very large, with new stakeholder entrants dynamically joining and leaving. Hence, buyers and sellers will be concerned about the trustworthiness of the sellers and the realized composite services as well as the trustworthiness of buyers and their behaviors. Current approaches to trust in networking are typically based on measures such as reputation, in a manner that will not scale to such large and dynamic ecosystems and does not encourage new entrants to join. We propose an architecture for intent-based 6G multi-stakeholder marketplaces inspired by decentralized finance (DeFi); namely, intent-based cryptocurrency swaps on blockchains/cross-chains. In such DeFi paradigms, trust is implicitly modeled via techno-economic risk management, rather than explicitly computed measures such as reputation. We propose how to leverage such DeFi paradigms to enable trustworthy intent-based multi-stakeholder 6G services marketplaces with SLAs, in a manner that encourages new entrants. Our approach is based on techno-economic risk vs. market-driven rewards as a proxy for trust, realized through intent-based blockchain/crosschain architectures with SLA assurance. Our work provides an enabler for establishing trusted multi-stakeholder interactions with UNEXT, an intelligent networking platform being created by Nokia Bell Labs.
As a data-free knowledge transfer paradigm, federated learning (FL) provides a novel solution for knowledge fusion in smart cities, especially in the field of Internet of Vehicles (IoV). However, the bandwidth bottleneck in the IoV limits the efficiency of federated collaboration, while trust issues associated with aggregation servers reduce users’ willingness to collaborate. To address these challenges, this article proposes a multiple collaborative iterative distillation-based swarm learning (CISL) framework for IoV. CISL leverages multiple collaborative iterative distillations to transform federated collaboration into serverless cross-device and cross-decentralized autonomous organization (DAO) knowledge transfer and fusion, enabling trustworthy swarm collaboration under bandwidth-constrained conditions. Moreover, it adaptively adjusts the inheritance and elimination of shared knowledge (SK) to enhance model adaptability and improve single-vehicle performance. Specifically, CISL proposes a collaborative iterative distillation mechanism that progressively integrates knowledge of other vehicles within the DAO, achieving cross-device SK fusion. Meanwhile, CISL introduces a multisage collaborative distillation mechanism, enabling each DAO to collaboratively distill and integrate SK from other DAOs, thereby expanding its knowledge domain. Additionally, CISL employs a dynamic balancing strategy to adaptively regulate the inheritance and elimination of SK, optimizing local models and enhancing their performance. Comprehensive experiments conducted on six benchmarks across two scenarios demonstrate that, compared to state-of-the-art methods, CISL exhibits superior adaptability and robustness across different datasets and task scenarios.