<p>The IoT-WSNs face the challenge of intrusion detection in wireless sensor networks due to dynamic patterns of traffic, limited resources of nodes, threat to privacy and susceptibility to routing and learning level attacks. Current intrusion detection systems (IDS) are either centralized learning which compromises on privacy and scalability or use federated learning which is not robust enough to handle model poisoning, manipulation of trust and routing level attacks. Furthermore, blockchain-based IDS systems have significant latency and overhead that restrict their functionality in resource-limited WSN. These constraints indicate a critical breach when it comes to realizing the lightweight, privacy-preserving, and trust-aware intrusion detection with a realistic implementation cost. This paper will fill this gap by proposing a secure federated blockchain-enabled IDS model of IoT-WSNs called SecuMesh-Net. The framework combines a temporal anomaly detector, based on Autoencoder and GRU, with federated learning to train models when decentralized, data privacy through the use of differential privacy, and finally, a blockchain based on PBFT to allow trust auditing and enforce routing security. The smart contracts provide trust- and energy-aware routing by isolating malicious nodes in real time and maintain the stability of the network. SecuMesh-Net is novel in its integrated approach to the time anomaly modelling, federated learning resilience, and lightweight blockchain auditing, separating consensus and operations of the IDS that require low latency. Comprehensive experiments based on real intrusion traces and traffic generated on NS-3 show that SecuMesh-Net has 97.4% detection, 2.1% false positive and 61 ms average latency, and per-event energy use of 2.4&#xa0;mJ. These findings affirm the applicability, effectiveness, and suitability of the framework to practical uses of the IoT-WSN.</p>