As more and more electronic devices get connected together, whole systems get more complicated in terms of development, maintenance and usage. New principles in design are needed to cope with this complexity. That is why we developed the artificial hormone system (AHS). The AHS enables an autonomous and decentralized task-allocation among a set of processing elements (PEs). The AHS works good in small-scaled nonhierarchical scenarios. For larger scaled and hierarchical scenarios, we developed the hierarchical artificial hormone system (HAHS) and presented a first abstract concept of the recursive artificial hormone system (RAHS). Both utilize the AHS in isolated smaller clusters of PEs to achieve the same functionality as the AHS with less communication. Thus, the whole task-set will be split into smaller and disjoint task-subsets which then will be distributed to the clusters. The HAHS uses a two-level hierarchy with a second regulation cycle between all clusters. The RAHS on the other hand is designed to manage a n-level hierarchy. In this article, we will present the first two implemented versions of the RAHS, their evaluation, and the comparison to the original AHS.

中文翻译：

---

大规模分层多核拓扑中的任务分配

随着越来越多的电子设备连接在一起，整个系统在开发、维护和使用方面变得更加复杂。需要新的设计原则来应对这种复杂性。这就是我们开发人工激素系统 (AHS) 的原因。AHS 支持在一组处理元件 (PE) 之间进行自主和分散的任务分配。AHS 在小规模的非分层场景中效果很好。对于更大规模和分层的场景，我们开发了分层人工激素系统 (HAHS)并提出了递归人工激素系统 (RAHS)的第一个抽象概念. 两者都在隔离的较小 PE 集群中使用 AHS，以实现与 AHS 相同的功能，但通信较少。因此，整个任务集将被分成更小且不相交的任务子集，然后将这些子集分发到集群中。HAHS 使用两级层次结构，在所有集群之间有第二个调节周期。另一方面，RAHS 旨在管理n级层次结构。在本文中，我们将介绍 RAHS 的前两个实施版本、它们的评估以及与原始 AHS 的比较。