Main Article Content
Federated learning, which is a distributed machine learning approach for preserving privacy and is thus widely used in numerous privacy issue applications, is involved in the area where privacy is of higher importance. To protect the privacy of users' local gradients while conducting federated learning, elliptical curve cryptography with block chain-based federated learning (ECC-BFL) is proposed here.Considerable consideration is given to factors including categorization accuracy, running time, communication overhead, computation overhead, and transaction speed. The values for these parameters are compared to three established techniques: the Biparing Method (BM), the Homomorphic Cryptosystem (HC), and the Multiple Authorities with Attribute-Based Signature scheme (MA-ABS). Additionally, a proposed Elliptical Curve Cryptography with Blockchain-based Federated Learning (ECC-BFL)technique is also considered.The suggested ECC-BFL was able to accomplish 95% classification accuracy, 65 sec. of operating time, 76% communication overhead, 63% calculation overhead, and 92% transaction speed.