Software systems increasingly interact with each other, forming ecosystems. Cloud is one such ecosystem that has evolved and enabled other technologies like IoT and containers. Such systems are very complex and heterogeneous because their components can have diverse origins, functions, security policies, and communication protocols, which makes it difficult to comprehend, utilize and consequently secure them. Abstract architectural models can be used to handle this complexity and heterogeneity but there is lack of work on precise, implementation/vendor neutral and holistic models which represent ecosystem components and their mutual interactions. We attempted to find similarities in systems and generalize to create abstract models for adding security. We represented the ecosystem as a Reference architecture (RA) and the ecosystem units as patterns. We started with a pattern diagram which showed all the components involved along with their mutual interactions and dependencies. We added components to the already existent Cloud security RA (SRA). Containers, being relatively new virtualization technology, did not have a precise and holistic reference architecture. We have built a partial RA for containers by identifying and modeling components of the ecosystem. Container security issues were identified from the literature as well as analysis of our patterns. We added corresponding security countermeasures to container RA as security patterns to build a container SRA. Finally, using container SRA as an example, we demonstrated an approach for RA validation. We have also built a composite pattern for fog computing that is an intermediate platform between Cloud and IoT devices. We represented an attack, Distributed Denial of Service (DDoS) using IoT devices, in the form of a misuse pattern which explains it from the attacker’s perspective. We found this modelbased approach useful to build RAs in a flexible and incremental way as components can be identified and added as the ecosystems expand. This provided us better insight to analyze security issues across boundaries of individual ecosystems. A unified, precise and holistic view of the system is not just useful for adding or evaluating security, this approach can also be used to ensure compliance, privacy, safety, reliability and/or governance for cloud and related ecosystems. This is the first work we know of where patterns and RAs are used to represent ecosystems and analyze their security. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection