The internet is getting credit for being a backbone of information sharing these days. As the demand of the transmission of information on the un-trusted network increases, people demanding the more secure mechanism to share data on hybrid networks. Now a day, the Ib contents contain multimedia which demands security along with a maximum compression ratio of transmission over the networks. The aim of this research paper is twofold, one the novel lightIight encryption method is proposed which is based on Permutation rounds with exclusive OR (P-XOR) and second the implementation of expert systems to auto detects three different security levels for hybrid devices/networks. P-XOR is a symmetric stream cipher with three keys of 128 bits along with five rounds. As per the need of heterogeneous devices and networks, three security levels are implemented on the compression stage of H.264 bit-streams. Three cipher algorithms, i.e. exclusive OR (XOR) for low level security, Advanced Encryption Standard (AES) for high level security and newly proposed P-XOR for medium level security are selected by the implementation of fuzzy rule based expert system. The proposed expert system intelligently selected the best security level by judging the device and network requirements for H.264 video transmission. The proposed framework has resolved the key problem of selecting the same level of hybrid infrastructure, i.e. the devices like mobiles, tablets, laptops and servers with differing requirements like energy, storage, and resolution and frame rate. The results on different sample videos have confirmed the validity of the proposed expert system for selection of different security levels. As a research contribution, the implementation of proposed framework has provided the minimum computational cost due to selective encryption, better performance in terms of selection time by the expert system, video format compliance and without network overhead for encrypted videos on said security levels.