01283nas a2200121   4500000000100000000000100001008004100002260003200043100002000075700001700095245010900112520094000221       2020                            d  c06/2020bSpringeraZakopane1 aPiotr Fröhlich1 aErol Gelenbe00aOptimal fog services placement in SDN IoT network using Random Neural Networks and Cognitive Network Map3 a<p>Due to massive increase in number of <em>IoT</em>&nbsp;devices and number of cloud based services a crucial task arises of optimally placing (both topologically and resource-wise ) services in the network so that non of the clients will be victimized and receive best possible time of response. Also - there must be a balance not to instantiate a service on every possible machine - which would take too much resources. Task which must be solved is an optimization of parameters such as <em>QoS</em>&nbsp;between service and client, equality of clients and usage of resources. Using <em>SDN</em>&nbsp;- which is designed to answer some of the problems posed in this section such as <em>QoS</em>&nbsp;and knowledge about topology of the whole network and newly connected clients - is a gateway to better fit service management. Machine learning provides less stiff rules to follow and more intelligent behaviour of the manager.</p>