OpenCLAW-Nexus: A Self-Reinforcing Trust Framework for Byzantine-Resilient Decentralized Federated Learning
2026-05-07T07:23:50Z•5b7b0df8711bff65dab05c34df08535a97ace259fa3e67da3451db8276fe90a4
AI training network resilienceBFT consensusByzantine resilienceLEO satellite routingLLM-based network agentsRDMA/MRC transportRL robustnessReGuardRep-FedAvgSADE troubleshooting policySRv6Sybil attackscamera-primed sensingdecentralized federated learningdifferential privacymmWave beamformingmulti-UAV IoVreputation systemsruntime protectionworst-case discovery
What happened
This collection of recent networking and ML-systems papers presents advances in resilience and performance at scale and highlights new/shifted attack surfaces. Key contributions: OpenCLAW-Nexus unifies reputation-based node selection, reputation-weighted aggregation (Rep-FedAvg), and reputation-aware BFT consensus to close the trust gap in fully decentralized federated learning — claiming formal separation of honest vs Byzantine under non-IID data and robustness to 20% Byzantines and large Sybil floods. MRC (an RDMA-based multipath transport) combined with static SRv6 routing and multi-plane/f
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_ni
- Record identifier
- 5b7b0df8711bff65dab05c34df08535a97ace259fa3e67da3451db8276fe90a4
- Enrichment time
- 2026-05-07T07:23:50Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.