These applications have high, ever-growing requirements in terms of security, reliability and performance. Currently, the development of these applications is heavily dependent on cloud, the abundant resources of which are a necessity for the computationally intensive Artificial Intelligence (AI) methods. However, cloud-native processing requires transmitting data between the end users and the cloud, which increases the latency, burdens the core network and raises privacy concerns. Hence, several computing paradigms, such as edge and fog computing, Multi-access Edge Computing (MEC) and cloudlets (Ren et al. (2020)), have emerged to bring the computing and storage resources from the cloud to the edge, closer to the end users. Even though these paradigms have differences in their architectural considerations and driving forces, they all have the same essence: placing and using computational resources between the end user and the distant cloud in order to reduce latency and energy consumption, as well as increase security and privacy by keeping the application data local. Bringing the intelligent applications onto the edge between the end users and the cloud is not a simple task. Traditional AI is inherently centralized and resource consuming, while the edge is inherently distributed and limited in resources. Further, the edge nodes are highly heterogeneous in terms of their capabilities, while the edge environment as a whole is characterized by intermittent connectivity, distributed and non-IID data, as well as geographically distributed, opportunistic computing resources (Kokkonen et al. (2022)). Research on developing and adapting AI methods to the edge environment has been coined as AI on Edge (Lovén et al. (2019); Deng et al. (2020)), which is an active research area with an ample amount of research (Deng et al. (2020); Xu et al. (2021); Park et al. (2021)).

In many self-organising systems the ability to extract necessary resources from the external environment is essential to the system's growth and survival. Examples include the extraction of sunlight and nutrients in organic plants, of monetary income in business organisations and of mobile robots in swarm intelligence actions. When operating within competitive, ever-changing environments, such systems must distribute their internal assets wisely so as to improve and adapt their ability to extract available resources. As the system size increases, the asset-distribution process often gets organised around a multi-scale control topology. This topology may be static (fixed) or dynamic (enabling growth and structural adaptation) depending on the system's internal constraints and adaptive mechanisms. In this paper, we expand on a plant-inspired asset-distribution model and introduce a more general multi-scale model applicable across a wider range of natural and artificial system domains. We study the impact that the topology of the multi-scale control process has upon the system's ability to self-adapt asset distribution when resource availability changes within the environment. Results show how different topological characteristics and different competition levels between system branches impact overall system profitability, adaptation delays and disturbances when environmental changes occur. These findings provide a basis for system designers to select the most suitable topology and configuration for their particular application and execution environment.