The increasing susceptibility of IoT devices to cyber threats underscores the limitations of conventional security measures in safeguarding the decentralized and resource-limited characteristics of IoT ecosystems. This study addresses the crucial problem of improving the efficiency and security of intrusion detection systems (IDSs) in IoT networks, all while ensuring data privacy is maintained. Prior studies have utilized a range of cryptographic methods to enhance the security of IoT systems. Nonetheless, numerous individuals continue to navigate the trade-off between strong protection and the constrained resources of IoT devices, especially within diverse and extensive networks. This study addresses this gap by introducing a hybrid blockchain-based framework that combines elliptic curve cryptography (ECC) with federated learning techniques. The framework utilizes a CNN-BiLSTM hybrid model, utilizing federated learning to enable decentralized and privacy-preserving threat detection across IoT devices. Zero-knowledge proofs (ZKPs) facilitate the authentication of security events without disclosing sensitive information, whereas Istanbul Byzantine Fault Tolerance (IBFT) guarantees dependable consensus in distributed networks. Performance evaluations demonstrate significant improvements in encryption and decryption durations, block generation, key production, throughput, and response times when juxtaposed with conventional cryptographic techniques. The proposed framework provides a scalable, secure, and effective approach to protect IoT environments from advancing cyber threats.