The Internet of Things (IoT) has revolutionized various industries by enabling data exchange between different devices across various domains such as smart cities, healthcare, industrial automation etc. However, managing access control with growing number of IoT devices brings major security challenges. Traditional access control mechanisms such as Role-Based Access Control(RBAC) and Attribute-Based Access Control(ABAC) become very complex and computationally expansive for the large scale iot networks. Besides these, Manufacturer Usage Description (MUD) based mechanism empowers networks to restrict IoT devices to communicate only with authorized endpoints, ensuring that each device sends and receives only the intended traffic while preventing unauthorized access or data transmission. However, the static MUD profiles provided by manufacturers are not adaptable to dynamic IoT environments, where devices frequently join, leave, or change behavior. Additionally, manually creating and updating MUD profiles may not be possible and prone to errors for dynamic and large scale IoT network. To address these limitations, this paper proposes an automated framework for generating and enforcing MUD profiles based on network behavior. The framework leverages the MUD specification by analyzing network traffic and extracting the most relevant features using mutual information (MI) scores. These features, which correlate strongly with device behavior, are then used in association rule mining (ARM) to generate refined access control rules. The rules are verified and integrated into the MUD profiles, ensuring automated policy enforcement. Furthermore, the MUD profiles are stored in a tamper-resistant manner using IPFS (InterPlanetary File System), preventing them from unauthorized modifications. The framework also utilizes smart contracts on a blockchain to verify and enforce security policies. The approach improves security by allowing only intended device interactions while denying abnormal traffic, and enhances performance through efficient rule generation and enforcement. The results demonstrate that the use of ARM with MI scores improves rule quality, reduces complexity, and facilitates faster, more reliable network operations in dynamic IoT environments.