00000nam 2200000zi 4500 9.881594 CaOODSP 20240219183525 m o d f cr cn||||||||| 191104t20132013quca ob f000 0 eng d CaOODSP eng rda CaOODSP eng eng fre D68-4/409-2013E-PDF Léchevin, N., author. Pattern recognition of socio-technical network vulnerabilities : modeling and preliminary results / Nicolas Léchevin, Anne-Laure Jousselme, Patrick Maupin, DRDC Valcartier. Quebec (Quebec) : Defence R&D Canada - Valcartier, 2013. ©2013 1 online resource (x, 60 pages, 2 unnumbered pages) : illustrations (chiefly colour). text txt rdacontent computer c rdamedia online resource cr rdacarrier Technical report ; DRDC Valcartier TR 2013-409 "December 2013." Includes bibliographical references (pages 49-52). "When faced with potentially disruptive events, the state of a network may unexpectedly evolve to regions of the state space where safe operating conditions are no longer ensured. It is thus highly desirable to relate the network characteristics and operating conditions to its vulnerabilities, if any, in order to mitigate risk expressed as a function of network inoperability and loss of quality of service. A pattern recognition approach is adopted to relate the structural features of the network to the loss of operating nodes and edges. Two types of networks are considered for analysis and simulation in this document. A network characterized by flow conservation and capacity constraints is adapted from a fuse model, which may lead, in some instances, to cascading events. A tactical swarm of robots is deployed either to achieve terrain surveillance coverage or to maintain client connectivity so that every client can communicate in remote area. In both cases, the swarm of robots should maintain its connectivity at each time instant. The swarm deployment adapts to the loss of a robot caused by such factors as hardware/software failure, enemy action, or the presence of malware. The motion strategy prioritizes the client coverage, which may entail possible losses of connectivity. Given the motion strategy at hand, the swarm presents vulnerabilities related to the loss of some nodes. The classifier, instrumental in performing pattern recognition, is trained from a sample of networks obtained by some probabilistic generator. The classifier is shown to model, and to some extent, predict quickly the vulnerabilities of a class of networks as a function of their structural properties"--Abstract, page i. Includes abstracts and summaries in English and French. Defence R&D Canada. Defence R&D Canada - Valcartier. Technical report (Defence R&D Canada) DRDC Valcartier TR 2013-409. (CaOODSP)9.820558 PDF 3.49 MB https://publications.gc.ca/collections/collection_2019/rddc-drdc/D68-4-409-2013-eng.pdf