This research will validate the premise of providing transport-as-a-service to networked applications in order to better facilitate their often disparate requirements. Instead of continuing to use the traditional, fixed-at-design-time transport layer approach used by almost all wide area network communicating devices, it will be shown that significant optimisation opportunities exist through the intelligent adaptation of an application's individual transport layer. Given that there are a large number of under-utilised, IETF-defined, transport protocols, there exists a relatively unexplored optimisation opportunity based on lever aging each protocol's unique advantages at the correct time. It will be shown that, through continuous, run-time, transport protocol selection and switching, it is possible to overcome the implicit multidimensional heterogeneity problems that researchers, system designers and operators face when developing systems to operate in resource-constrained Distributed Real time Embedded (DRE) environments. To evaluate the performance benefits and deployment considerations this approach can provide, a fully-operational software middleware system called DIRECTOR will be presented and tested using an emulated Smart Grid data network providing a real world target system. With the recent surge of distributed, machine-to-machine, networked devices that are generating large amounts of information-rich data in the form of the Internet of Things and the Smart Grid, a viable solution to the looming problems of facilitating efficient, large-scale DRE communication systems is required. This thesis will present a transparent middleware system, enabling applications to receive their specified optimal transport layer, as such a solution .