Articulation and Clarification of the Dendritic Cell Algorithm - Artificial Immune Systems - page 406

The characteristics of the relevant signals are summarised below:

- PAMPS are pre-defined bacterial signatures, causing the maturation of im-
mature DCs to mature DCs through expression of 'mature cytokines'.
- Danger signals are released as a result of damage to tissue cells, also increas-
ing mature DC cytokines, and have a lower potency than PAMPs.
- Safe signals are released as a result of regulated cell death and cause an
increase in semi-mature DC cytokines, and reduce the output of mature DC
cytokines
- Inflammatory cytokines are derived from general tissue distress and amplify
the effects of the other three signals but are not sucient to cause any effect
onimmatureDCswhenusedinisolation.

3 Dendritic Cells in silico

The Dendritic Cell Algorithm (DCA) was developed as part of the Danger
Project[1], which aims to find the missing link between AIS and Intrusion Detec-
tion through the application of the danger theory[8]. The danger theory proposes
that the immune system responds when damage to the host is detected, rather
than discriminating between self and non-self proteins. The project encompasses
artificial tissue[3] and T-cells[7], and the libtissue framework[11]. The DCs are
the detection component developed within this project.

3.1 Libtissue
Libtissue is a software system which allows the implementation and testing of
AIS algorithms on real-world problems based on principles of innate immunol-
ogy [10], [11]. It allows researchers to implement AIS algorithms as a collection
of cells, antigen and signals interacting within a tissue compartment. The im-
plementation has a client/server architecture, separating data collection from
data processing. Input data to the tissue compartment is generated by sensors
monitoring environmental, behavioural or context data through the libtissue
client, transforming this data into antigen and signals. AIS algorithms can be
implemented within the libtissue server, as libtissue provides a convenient
programming environment. Both client and server APIs allow new antigen and
signal sources to be added to libtissue servers, and the testing of the same
algorithm with a number of different data sources. Input data from the tissue
client is represented in a tissue compartment contained on the tissue server.
A tissue compartment is a space in which cells, signals and antigen interact.
Each tissue compartment has a fixed-size antigen store where antigen provided
by libtissue clients is placed. The tissue compartment also stores levels of
signals, set either by tissue clients or cells.

3.2 Abstract View of the DCA
The DCA is implemented as a libtissue tissue server. Input signals are com-
bined with a second source of data, such as a data item ID, or program ID