Reducing software assurance risks for security-critical and safety-critical systems

According to the Office of the Assistant Secretary of Defense for Research and Engineering (ASD(R&E)), the US Department of Defense (DoD) recognizes that there is a “persistent lack of a consistent approach ... for the certification of software assurance tools, testing and methodologies” [1]. As a result, the ASD(R&E) is seeking “to address vulnerabilities and weaknesses to cyber threats of the software that operates ... routine applications and critical kinetic systems ...” The mitigation of these risks has been recognized as a significant issue to be addressed in both the public and private sectors. In this paper we examine deficiencies in various software-assurance approaches and suggest ways in which they can be improved. We take a broad look at current approaches, identify their inherent weaknesses and propose approaches that serve to reduce risks. Some technical, economic and governance issues are: (1) Development of software-assurance technical standards (2) Management of software-assurance standards (3) Evaluation of tools, techniques, and metrics (4) Determination of update frequency for tools, techniques (5) Focus on most pressing threats to software systems (6) Suggestions as to risk-reducing research areas (7) Establishment of models of the economics of software-assurance solutions, and testing and certifying software We show that, in order to improve current software assurance policy and practices, particularly with respect to security, there has to be a major overhaul in how software is developed, especially with respect to the requirements and testing phases of the SDLC (Software Development Lifecycle). We also suggest that the current preventative approaches are inadequate and that greater reliance should be placed upon avoidance and deterrence. We also recommend that those developing and operating security-critical and safety-critical systems exchange best-ofbreed software assurance methods to prevent the vulnerability of components leading to compromise of entire systems of systems. The recent catastrophic loss of a Malaysia Airlines airplane is then presented as an example of possible compromises of physical and logical security of on-board communications and management and control systems.