The client authentication is a significant process in clientserver systems. Such a process is highly secure when a client may be authenticated according to a set of unique verifiable data, i.e., biometric traits. However, biometric based systems with the low-cost, dense biometric sensors, and power of fast processing need a method of automatic client recognition for the robust client authentication. Such a method faces three challenges: (1) the effective recognition of the biometric patterns inputted to the system, (2) the provision of security to prevent the vulnerability of the system, and (3) the preparation of personal privacy. Many remote biometric authentication schemes have been developed to establish secure mutual communication between a client as a device node and server over an untrusted channel. By employing a secure remote biometric based authentication protocol, a client that acts in a node and a server that contains resources can authenticate each other in a secure and trustable manner. In our previous work, we proposed a digest based authentication method that preserves privacy of clients biometric templates and authenticates the client securely by generating non-deterministic semi-digest. By reviewing and cryptanalyzing this method, in the current paper, we focus on the improvement of the method for providing the invulnerability against user anonymity and server masquerading attacks. We show that our improved scheme is secure against the attacks and prove its functionality features. (C) 2019 Elsevier B.V. All rights reserved.