Through the advent of the smartphone and other enabling wireless technologies there are now increased interaction between mobile devices, the physical environment and data sources within it. This scenario is known as the converging Cyber-Physical World (CPW) [Conti et al. 2012] and within this, opportunistic networking [Boldrini and Passarella 2013] is an important enabling paradigm. Opportunistic networking is an entirely self organised form of communication which functions by mobile devices, such as smartphones, transiently connecting when they come into range. This is known as the store-carry and forward paradigm, which opens up new ways to create and share knowledge through data dissemination. This makes them ideal for crowdsourcing applications where distributed resources are harnessed to provide new services for a wide range of emerging applications including smart cities, e-health, intelligent transportation systems [Conti et al. 2012]. Unlike many forms of online crowdsourcing, opportunistic networking differs in that the providers of resources are also the consumers of sub-services, described by the subset of data that is relevant for their needs and interests. This participatory prosumer model is a distinctive feature of opportunistic networking. The concept of opportunistic networking brings networking closer to the disposition of the human user because the mobile devices such as smartphones which perform the networking function are carried around throughout the users day-to-day activity. Due to this these devices can act as cyber-physical proxies for their human users [Whitaker et al. 2015], potentially autonomously discovering, collecting and evaluating data from

Modern information technology services largely depend on cloud infrastructures to provide their services. These cloud infrastructures are built on top of datacenter networks (DCNs) constructed with high-speed links, fast switching gear, and redundancy to offer better flexibility and resiliency. In this environment, network traffic includes long-lived (elephant) and short-lived (mice) flows with partitioned and aggregated traffic patterns. Although SDN-based approaches can efficiently allocate networking resources for such flows, the overhead due to network reconfiguration can be significant. With limited capacity of Ternary Content-Addressable Memory (TCAM) deployed in an OpenFlow enabled switch, it is crucial to determine which forwarding rules should remain in the flow table, and which rules should be processed by the SDN controller in case of a table-miss on the SDN switch. This is needed in order to obtain the flow entries that satisfy the goal of reducing the long-term control plane overhead introduced between the controller and the switches. To achieve this goal, we propose a machine learning technique that utilizes two variations of reinforcement learning (RL) algorithms-the first of which is traditional reinforcement learning algorithm based while the other is deep reinforcement learning based. Emulation results using the RL algorithm show around 60% improvement in reducing the long-term control plane overhead, and around 14% improvement in the table-hit ratio compared to the Multiple Bloom Filters (MBF) method given a fixed size flow table of 4KB.