The process of Federated learning (FL) allows different devices to jointly create a model through collaboration without exchanging training information. FL protects diversified applications from sensitive data leakage through mechanics including differential privacy and homomorphic encryption and secure multi-party computation which meet privacy legislation. FL serves diverse healthcare and financial sectors as well as IoT operations alongside edge computing applications because it keeps sensitive information within individual devices. As much as FL provides benefits its implementation faces three main challenges related to communication overhead and non-IID data distributions together with security vulnerabilities.