Bidirectional wireless communication is employed in various smart grid components such as smart meters and control and monitoring applications where security is vital. The Trusted Third Party (TTP) and wireless connectivity between the smart meter and the third party in the key management-based encryption techniques for the smart grid are expected to be totally trustworthy and dependable. In a wired/wireless medium, however, a man-in-the-middle may seek to disrupt, monitor and manipulate the network, or simply execute a replay attack, revealing its vulnerability. Recognizing this, this study presents a novel authentication management (model) comprised of two layer security schema. The first layer implements an efficient novel encryption method for secure data exchange between meters and control center with the help of two partially trusted simple servers (constitutes the TTP). In this setting, one server handles the data encryption between the meter and control center/central database, and the other server administers the random sequence of data transmission. The second layer monitors and verifies exchanged data packets among smart meters. It detects abnormal packets from suspicious sources. To implement this node-to-node authentication, One class support vector machine algorithm is proposed which takes advantages of the location information as well as the data transmission history (node identification, packet size, and data transmission frequency). This schema secures data communication, and imposes a comprehensive privacy throughout the system without considerably extending the complexity of the conventional key management scheme.