The general concept of security in message communications may be traced to the advent of human civilization. In contrast, however, security in automation and control is a recent phenomenon, originating with the computer age and is rapidly gaining importance with the proliferation of networks. With computer networks integrating the dual functions of (i) communications and (ii) automation and control, computer network security must address the security issues inherent in both communications and automation. Until recently, research and development in computer security was strongly linked with cryptography including encryption and decryption of electronic messages. However, as computer networks have started to proliferate into large, complex, real-world systems, such as electronic banking, the power grid, and the proposed intelligent vehicle highway system, the authors believe that computer network security has transcended the traditional definition and has migrated to a higher, logical level. In current and future networks, the information riding on the network may control parts of the network while the control, in turn, may ensure the correct propagation of information from the source to the intended destination. Thus, networks constitute complex, multi-dimensional entities that require security at different levels of both network hardware and network software. This paper first reviews the current literature in network security, in detail. Second, it presents the definition of the network rating model, arrived at by consensus, at the National Security Agency's Network Rating Model (NRM) conference in March 1996 and a subsequent NRM author's group workshop in July 1996. Third, it proposes a fundamental framework for network security that consists of eight perspectives of network security and nine attributes of a secure network. The framework approach was adopted at the NRM author's group workshop. The perspectives, termed «pillars» in this paper, individually provide orthogonal views of network security and collectively constitute a comprehensive stable structure that supports the total network security. The attributes refer to the inherent characteristics of a secure network. The paper then presents a list of the research efforts in high-level security and intrusion resistance and logically argues for their need towards comprehensive network security. A few of the key research areas stressed in this position paper include, (i) the stability of networks under perturbation, (ii) formal proofs of correctness of the algorithms designed to ensure network security, (iii) the use of fuzzy logic and fuzzy thresholds for modeling networks and security criteria, (iv) dynamicity of the parameters constituting the security criteria, over time, to foil intrusion attempts, and (v) the design of innovative schemes aimed at reducing performance impact of security overhead by exploiting key environment or protocol characteristics of the network under consideration.