This paper presents a critical review and discussion of previous works in evolvable hardware (EH) and the research directions in this field. According to the authors, they “highlight some of the successes” but “also [consider] why the field is far from reaching these goals” (that is, achieving success and being competitive with human designs).

Initially, the authors try to define the main concepts, specifying and differentiating two subfields: evolvable hardware design and adaptive hardware. They define what is and what is not evolvable hardware, as well as the main platforms and devices studied, considering both analog and digital devices and also intrinsic (on-chip) and extrinsic (simulated) implementations. They then present a limited comparison of some proposed evolvable hardware systems according to their properties and implementations, followed by a few success stories. They propose a discussion about the challenges in this field and pose some high-level and philosophical questions about the difficulties and problems in the development of evolvable hardware, which are somewhat questionable. Finally, they turn to a brief discussion of practical aspects, such as scalability, optimization, search space, measures, fitness function, and fitness evaluation. They conclude with a discussion of new approaches.

This is a good review of this research field, which is neither complete nor conclusive; the references are interesting, as are some posed questions and discussions.