Cloud computing is one of the latest technologies that is much talked about these days. In the IT world cloud computing is considered as a revolutionary technology which has been built by consolidating the already existing techniques such as grid computing, virtualization, and cryptography. Apart from having a number of advantages, cloud computing is still struggling with some security issues. System security is one of the major roadblocks for the sustainability of cloud in the future. Since the introduction of cloud in the current world the work is continuously being done for the security issues. A number of techniques have already been deployed to overcome these security concerns and still people are working to improve the security of cloud based systems. These schemes are mainly classified as Access control policies and Attribute based Encryption schemes. But all these schemes are still facing the problem of expressibility. One more constraint is that they only work for systems where the data owner and the service provider are in the same trusted domain. Apart from this these policies are also short of flexibility in attribute management and lacks scalability in dealing with multiple-levels of attribute authorities. The primary focus of our work is to overcome these limitations in the policies. Our aim is to provide securable and flexible access control which will be authenticated properly. This will help to maintain the data securely. In this paper we proposed a ML-ASBE scheme which will overcome all the limitations of the existing system. This will also deal with collusion resistance and decreases the complexity of computation.