Toward Adaptive and Reflective Middleware for Network - Centric Combat Systems

N ew and planned Department of Defense (DoD) combat systems are inherently network-centric, distributed real-time and embedded (DRE) systems of systems. Combat systems have historically been developed via multiple technology bases, where each system brings its own networks, computers, displays, software, and people to maintain and operate it. Unfortunately, not only are these stove-pipe architectures proprietary, but by tightly coupling many functional and quality of service (QoS) aspects they impede these DRE system features: • Assurability is needed to guarantee efficient , predictable, scalable, and dependable QoS from sensors to shooters. • Adaptability is needed to (re)configure combat systems dynamically to support varying workloads or missions over their life cycles. • Affordability is needed to reduce initial nonrecurring combat system acquisition costs and recurring upgrade and evolution costs. In recognition of the importance of enhancing affordability, recent DoD programs such as the Aegis destroyer program [1], the New Attack Submarine program [2], the Weapons Systems Open Architecture program [3], and the Unmanned Combat Air Vehicle (UCAV) program [4] have adopted strong open systems approaches to system design and commercial-off-the-shelf (COTS) refresh strategies. Ultimately, open systems approaches are more likely to be robust with respect to change over the long life cycles typical of military systems. For example, the affordability of certain types of DoD systems such as logistics and mission planning have been improved by using COTS technologies. However, many of today's procurement efforts aimed at integrating COTS into mission-critical DRE combat systems have largely failed to support life-cycle affordability, assurability, and adaptability effectively since they focus mainly on initial nonrecurring acquisition costs and do not reduce recurring software life-cycle costs, such as COTS refresh and subset-ting combat systems for foreign military sales [5]. Likewise, many COTS products lack support for controlling key QoS properties such as predictable latency, jit-ter, and throughput; scalability; dependability ; and security. The inability to control these QoS properties with sufficient confidence compromises combat system adaptability and assurability, e.g., a perturbation in the behavior of a COTS product that would be acceptable in commercial applications could lead to loss of life and property in military applications. Historically, conventional COTS software has been unsuitable for use in mission critical DRE combat systems due to either of the following: • It is flexible and standard, but incapable of guaranteeing stringent QoS demands, which restricts assurability. • It is partially QoS-enabled, but inflexible and non-standard, which restricts adaptability …