Many　communication　systems,　e.　g.,　internet,　can　be　modeled　as　complex　networks.　For　such　networks,　immunization　strategies　are　necessary　for　preventing　malicious　attacks　or　viruses　being　percolated　from　a　node　to　its　neighboring　nodes　following　their　connectivities.　In　recent　years,　various　immunization　strategies　have　been　proposed　and　demonstrated,　most　of　which　rest　on　the　assumptions　that　the　strategies　can　be　executed　in　a　centralized　manner　and/or　that　the　complex　network　at　hand　is　reasonably　stable　(its　topology　will　not　change　overtime).　In　other　words,　it　would　be　difficult　to　apply　them　in　a　decentralized　network　environment,　as　often　found　in　the　real　world.　In　this　paper,　we　propose　a　decentralized　and　scalable　immunization　strategy　based　on　a　self-organized　computing　approach　called　autonomy-oriented　computing　(AOC)　[1],　[2].　In　this　strategy,　autonomous　behavior-based　entities　are　deployed　in　a　decentralized　network,　and　are　capable　of　collectively　finding　those　nodes　with　high　degrees　of　conductivities　(i.e.,　those　that　can　readily　spread　viruses).　Through　experiments　involving　both　synthetic　and　real-world　networks,　we　demonstrate　that　this　strategy　can　effectively　and　efficiently　locate　highly-connected　nodes　in　decentralized　complex　network　environments　of　various　topologies,　and　it　is　also　scalable　in　handling　large-scale　decentralized　networks.　We　have　compared　our　strategy　with　some　of　the　well-known　strategies,　including　acquaintance　and　covering　strategies　on　both　synthetic　and　real-world　networks.