Abstract
This paper focuses on the resilience of a nature-inspired class of algorithms. The issues related to resilience fall under a very wide umbrella. The uncertainties that we face in the world require the need of resilient systems in all domains. Software resilience is certainly of critical importance, due to the presence of software applications which are embedded in numerous operational and strategic systems. For Ant Colony Optimization (ACO), one of the most successful heuristic methods inspired by the communication processes in entomology, performance and convergence issues have been intensively studied by the scientific community. Our approach addresses the resilience of MAX–MIN Ant System (MMAS), one of the most efficient ACO algorithms, when studied in relation with Traveling Salesman Problem (TSP). We introduce a set of parameters that allow the management of real-life situations, such as imprecise or missing data and disturbances in the regular computing process. Several metrics are involved, and a statistical analysis is performed. The resilience of the adapted MMAS is analyzed and discussed. A broad outline on future research directions is given in connection with new trends concerning the design of resilient systems.
Highlights
In recent decades, the community of researchers from various domains has shown a growing interest towards systems resilience
A shift from robustness-centered design to principles of more flexible and adaptive design has been noticed [33]. This approach includes the following ideas: composing the measures of different aspects in order to reason about resilience; the metrics should be particular for a specific perturbation; the metrics should be dependent on the system boundaries; the customer requirements drive the metrics for resilience; take into consideration other aspects except for the system output
For the original MAX–MIN Ant System (MMAS), recovery speed ∈[0.4934, 1.6154], while for the applications corresponding to data in Table 1, recovery speed ∈ [0.4173, 1.3642] and only three values are smaller than the minimum recovery speed for MMAS with ampl = 0
Summary
The community of researchers from various domains has shown a growing interest towards systems resilience. The list of domains where systems resilience is important is long, and specific definitions have been provided: engineering [1], economics [2,3], environment [4,5], ecology [6,7], psychology and neurobiology [8,9,10], sociology [11]. Given the wide interest and importance of the concept, for researchers but for policymakers too, numerous and sometimes diverging interpretations and perceptions have been proposed for Mathematics 2020, 8, 752; doi:10.3390/math8050752 www.mdpi.com/journal/mathematics. Considerations on resilience, as it was defined and approached by researchers in various domains of human activity, are given, which bridges this concept with that of algorithmic performance.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
