In this sub-project, we introduce self-management techniques for the self-management and –regulation of an artificial immune system using lots of artificial cells. These artificial cells perform autonomously basic tasks, use basic communication techniques, and have knowledge about the local situation around the cell. The system reaches the global goals because each artificial cell reaches the basic goals.
One problem in such a Complex Adaptive System (CAS) is that there is not any guarantee that the system reaches the goals. Especially in Network Security, it is essential that there is a certain amount of security in each node. Furthermore, the artificial cells require “basic” information where, when, and which process is to be done in order to secure the network.
Currently, we introduced the following techniques for the self-management of an artificial immune system like SANA:
Idea:
A node notifies that it does not have enough artificial cells/the security level is low and this notification is distributed to an area of the network. Artificial cells receive this notification and decide autonomously whether they move towards this node or not; the artificial cells must not move there because the artificial cells are autonomous and behave independent from the network.
Performance:
The approach works well and the probability that cells move to a node with insufficient security level is higher than if the artificial cells does not receive the notification. Intuitively this is clear because the artificial cells receive more information and they can react using this information.
For more details of this approach, please visit
Self-Management Notification Approach in Detail
Trace of moving artificial cells so that an artificial cell can decide where to go next
In
SANA and other CAS are lots of specialized, mobile, and autonomous artificial cells. These cells move through the network and perform certain tasks. Furthermore, there are tasks, which must be done regularly but not every time step, e.g. monitoring, status checking, or data collecting. Thus, an artificial cell should see if another artificial cell of the same type already done this task a few time steps before.
For this problem, we introduce another self-management technique in SANA. An artificial cell of the type explained above releases a trace that disappears over time. Using receptors, this trace is assigned to a specific type of artificial cells. Thus, other artificial cells of this type receive information either that there were no artificial cell on this way in the last short-time or when there was the last artificial cell of this type.