Scholarly article on topic 'Modeling of Biological Immune System Mapped on Situation Awareness Model'

Modeling of Biological Immune System Mapped on Situation Awareness Model Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Computer Science
OECD Field of science
Keywords
{Antibody / "Biological Immune System" / "Endsley Model of Situation Awareness" / "Situation Awareness"}

Abstract of research paper on Materials engineering, author of scientific article — Touraj Banirostam, Hossein Parvar

Abstract Situation Awareness (SA) is a fundamental characteristic for a system. For understanding the effective parameters in SA, by describing the Biological Immune System, a model for behavior of antibody are proposed. Furthermore, four levels of Endsley Model of SA (EMSA) will be described and SA process in the proposed model will compared and mapped on EMSA. For simulating the proposed model a notation for agents are introduced and some parameters of two main categories in SA are presented. By running the simulation in different cases, the effect of memory, learning ability and communication in SA of each agent will be illustrated and compared. The results emphasize on importance of learning and communication in SA and robustness of a system. By recognizing the important parameters of SA, designing and implementation of self-aware system would be possible.

Academic research paper on topic "Modeling of Biological Immune System Mapped on Situation Awareness Model"

Available online at www.sciencedirect.com

SciVerse ScienceDirect

Procedía Computer Science 19 (2013) 1088 - 1093

The 8th International Symposium on Intelligent Systems Techniques for Ad hoc and Wireless Sensor Networks (IST-AWSN)

Modeling of Biological Immune System Mapped on Situation Awareness Model

Touraj Banirostama* and Hossein Parvarb

aComputer Dep., Islamic Azad University, Central Tehran Branch, Hamila Blov., Pounak, Tehran, Iran bComputer Dep., Science and Research Branch, Islamic Azad University, Pounak, Tehran , Iran

Abstract

Situation Awareness (SA) is a fundamental characteristic for a system. For understanding the effective parameters in SA, by describing the Biological Immune System, a model for behavior of antibody are proposed. Furthermore, four levels of Endsley Model of SA (EMSA) will be described and SA process in the proposed model will compared and mapped on EMSA. For simulating the proposed model a notation for agents are introduced and some parameters of two main categories in SA are presented. By running the simulation in different cases, the effect of memory, learning ability and communication in SA of each agent will be illustrated and compared. The results emphasize on importance of learning and communication in SA and robustness of a system. By recognizing the important parameters of SA, designing and implementation of self-aware system would be possible.

© 2013 The Authors.yPublished by Elsevier B.V.

Selection and peer-review under responsibility of Elhadi M. Shakshuki

Keywors: Antibody; Biological Immune System; Endsley Model of Situation Awareness; Situation Awareness

1. Main text

In many complex system [1] applications such as emergency management, disaster management, crisis management, and logistics management, Situation Awareness (SA) is required to make an effective decision. The more awareness is obtained the more effective decision. Therefore, the decision maker in these systems may be overwhelmed with SA [2 and 3]. As Adhoc and Wireless Sensor Network (WSN) are distributed systems, better SA in the elements of these systems caused better perception and consciousness about the environment. Therefore, the system will be self-aware WSN and reaction of the system will be effective and efficient. Based on Endsley's definition [4], S A is "the perception of

* Corresponding author. Tel.: +98-912-3587515; fax: +98-21-44227816. E-mail address: banirostan@iauctb.ac.ir, banirostam@yahoo.com

1877-0509 © 2013 The Authors. Published by Elsevier B.V. Selection and peer-review under responsibility of Elhadi M. Shakshuki doi:10.1016/j.procs.2013.06.153

environmental elements with respect to time and/or space, the comprehension of their meaning, and the projection of their status after some variable has changed, such as time, or some other variable, such as a predetermined event".

By comprehension of natural world new idea and approaches would be presented in science. Inspiring from biological systems, new and intelligent approaches could be introduced in different science [5]. One of the intelligent biological phenomena is the Biological Immune System (BIS). Antibodies in the BIS like elements in WSN use SA for confronting against antigens.

In the following section the concept of SA and EMSA will be described briefly, then the BIS structure will be considered shortly and behavioral model of an antibody proposed. Furthermore, the proposed model compares and maps on EMSA in section 3. Therefore, the effective parameters of SA in the BIS will be recognized. Finding out the effective parameters in SA based on the BIS could be useful for designing and implementation of self-aware system. Through description and proposed model of the BIS, a notation for simulating the behavior of an antibody will be presented in section 4. In section 5 the results of simulation will be illustrated and finally, conclusions will end the paper.

2. Situation Awareness (SA)

Endsley proposed a model for SA that is considered at three levels of perception, comprehension and projection [6]. At the first level, perception of objects and events, states and values are obtained through sensing, detecting and identifying. At the second level, comprehension, the meaning of the critical factors obtained at the perception level, implications and the types of situations are specified. The meaning of factors should be understood based on the operator's goals. Furthermore, interpretation and synthesis also occur at this level. Finally, the third level, projection, projects future scenarios and the possible outcomes and results. Prediction and simulation occur at this level. Later on a fourth level named resolution, was added to this model. The intention, the course of action and cooperation occur at this level that also includes planning and decision making.

In this model, in addition to the environment, system factors and individual factors also affect on SA. When individual tasks are increased and processing load is restricted, the process of SA is slowed down. Therefore, individual capabilities, experiences and exercises which are stored in the long term and working memory, increase the performance of individuals and decrease the process load.

3. The Biological Immune System

Antibody as one of the main elements in the BIS can recognize self (body cells-antibody) and non-self (all other elements-antigen). More details about behavior of the BIS and antibodies are presented in [5 and 7]. There are many models for BIS behavior. For example, in [8] the BIS modeled as Biological Complex Adaptive System and in [9] a formal definition of antibody, antigen and a theoretical model of the BIS is presented. More models, their application and structure are considered in [5] by details.

3.1. An Antibody Behavior Model

In the BIS, antibodies can memorize the structure of some antigens and just only recognize and confront the antigens which its structure has been memorized. Each antibody has two different receptors; Paratope uses for antigens recognition and Idiotope for communicate between antibodies. Furthermore, the part of antigens that recognized by antibodies called Epitope [5 and 7]. Figure 1 shows the different parts of an antibody.

Every antibody monitors its environment and determines self and non-self elements. The non-self structure would be checked by the antibody's memory. If they matched, the antibody would be able to

confront that antigen. Else, the antibody tries to adapt him with the structure of antigen and learn its structure. If the antibody had the ability to confront, it would migrate to the lymph nodes to start autopoiesis [8] and if it was not able, the antibody would keep on moving in the environment.

Fig. 1 .Different parts of an antibody

Figure 2 shows the proposed model of an antibody's behavior. Based on proposed model, behavior (internal process) of each antibody could be state in four steps of Monitoring, Analyse, Planning and Execution. The memory would be used in the last three steps. If an antibody considered as an agent, the steps could be described as below [8]:

• Monitoring: randomly movement and monitor environment for finding agents are exist in environment.

• Analyse: recognizing self and non-self elements based on memorized structures.

• Planning: finding a non-self element. In this case, the self agent adds non-self agent's structure into memory (learning) and could interact with other self agents.

• Execute: depends on the decision making in planning step antibody may migrate to lymph nodes, reproduce itself and finally attack the antigens, or communicate with other antibodies.

Fig. 2. Behavioral model of an antibody

Within the step of execution, the behavior has been applied to the environment and the feedback lead to evolution in the agents. Through the feedback learning process is occurred.

3.2. Mapping of Proposed Model on EMSA

Fist level of SA is Perception of the elements in the environment. It means the first step in achieving SA involves perceiving the status, attributes, and dynamics of relevant elements in the environment. This level of SA is obtained by the Monitor step in proposed model. Antibodies by using Paratope and Idiotope can achieve the perception of the environnement.

Second level of SA is Comprehension of the current situation. It is based on a synthesis of disjointed fist level elements. The second level of SA goes beyond simply being aware of the elements that are present to include an understanding of the significance of those elements in light of the specific goal. The

tasks of this level are achieved in analysed step of proposed model. Antibodies by matching of Paratope with Epitope of antigen and also using Idiotope network [5] of antibodies achieve the comprehension.

Third level of SA is Projection of future status. It is the ability to project the future actions of the elements in the environment, at least in the near term. This is achieved through knowledge of the status and dynamics of the elements and a comprehension of the situation (both first and second levels). The third level is equal with a part of planning step in proposed model. Antibodies by adaptation, migration and autopoiesis achieve projection.

Forth level of SA is Resolution means planning and decision making. The forth level is equal to a part of planning and execution in proposed model. Also, communication is happen in this level.

4. The BIS Simulation

For simulating the proposed model a formal structure like [9] is required. Two groups of agents -antibody (self) and antigen (non-self) - has been used for simulation. Eqs. (1) and (2) describe them [8].

AgentSeif = y(.Aw, ID, Co, Mo, Me, En) (1)

AgentNonSelf = y(Aw> Co> Mo> En) (2)

Where Aw is the perception of an agent about the environment every moment; ID is agent identification code; Co is agent coordination; Mo is agent direction in next movement; En is an agent exist energy and Me in antibody agents shows the memory of an antibody agent. Each antibody agent's memory includes a determined number of antigen agents' code initiate randomly and /is a function between the mentioned parameters. Some main characteristics of the antibody agent are:

• Learning: Antibody agents are able to add new recognized codes to its memory. The process of learning happens when an agent adds new recognized antigen's code into its memory. As time goes on, this process results evolution of the agent and therefore, makes the agent show a better reaction in the next confrontations. In simulation program, when an antibody recognized an antigen measures minimum Hamming Distance [5] between antigen's code and existing code in its memory based on Eq. (3). Eq.

(4) expresses learning of the antigen's code.

mHD^nt = Comp[MinHD(AgentNonself,Agent£elf)] (3)

, . Agent _ {insert(Aaent'^onSelf) to Agent%*lf if mHDstsh

LearningSelf - {No 0peratlon if mHD>tsh (4)

Where in Eq. (3) Comp is computation of minimum Hamming distance (mHD) between the ID of non-self agent and the memory of self agent. In Eq. (4) tsh is the threshold value that adjusted by the user.

• Communication: New recognized codes have been sent to the other agents through Idiotypic network

[5]. Agents by learning and communication have SA and could be resulted data fusion in antibodies.

Communication^^ : if (CoAgentl = CoAgent2) then Swap (AgentiDetection,Agent2Detection) (5)

Where, AgentDetectlon indicates the code of antigen recognized by antibody agent and saved in antibody short term memory.

5. Results of Simulation

A Multi Agent System has been designed and simulated in Netlogo environment for simulating the BIS. Two groups of agents - antibody and antigen- have been used in simulation in a way which they act against each other. They obey figure 2 for behavior generation. Antibody agents have same ID code and know each other but antigen agents have different random ID code. An antigen recognizing by an antibody is occurred in analyses step of proposed model (second level of SA). The way of recognition is SA. Antibody

agents would not know all antigen agents because they only have a memory with determined random codes of antigen agents (three codes in initial state).

Based on Endsley's model of SA, there are two main factors effect on SA. The first one is Core of SA that indicates four defined levels in section 2 and 3.2 and the second one is effective parameters like system capabilities, stress and work load, memory, experience, goals, aims and learning. For considering the effect of second part in SA the effect of learning and memory are considered in simulation of the BIS. Based on Eq. (3) and (4) two different ability of learning have been defined for antibody agents. In the low learning ability, the amount of threshold is 3 and antibody agents can only recognize unknown antigen with high similarity with the known ones. But in high learning ability, the amount of threshold is 9 and antibody agents recognize antigens which minimum Hamming Distance of their code with the existed codes in the memory of antibody agents are equal or less than 9. Furthermore, two different levels of initiate memory are defined for antibody agents. In low memory size, the memory initiates with 3 random codes of antigen and in high memory size it initiates with 9 random codes. Two of four different states are chosen and described in table 1, for considering in simulation program.

Table l.Two different stats of antibody's ability

Learning Ability Memory Size State A 3 9

State B 9 3

For considering the effect of parameters, robustness time in antibody as a benchmark is chosen. The robustness time for each groups of agents is the time which at least one agent with the ability of autopoiesis can stay alive. In each state (A and B) the simulation has been run for 100 times, and the process leads to the below results. As illustrated in Fig. 3-A, in the state A more data in the initial state causes antibody agents don't need to recognize antigen ones. In state B, because of increasing learning ability and decreasing memory, antibody agents could recognize more antigen agents than in state A.

Fig. 3.Average of learning in each state

In figure 3, X axis indicates the state and in figure 3-A, Y axis indicates number of learning. In figure 3-B, Y axis indicates percentage of robustness in 100 times running. As it is evidence a system with more ability of learning and less memory is more robust than a system with less learning and more memory.

In addition, as shown in figure 4, in same situation antibodies with communication and learning are more robust than without it. The results show memory, learning and communication, are important factors in SA of the elements of a system. Furthermore, it would be realized elements of a system with ability of learn and communication achieve SA in less time and become more robust. Therefore, by designing

WSNs based on agent with learning and communication ability the process of SA needs lees time and the system will be more robust. In figure 4, Y axis shows the average of robustness time for 100 running.

Fig. 4.Effect of learning and communication

6. Conclusions and Results

In this paper a model for antibody behavior was proposed. The proposed model was compared and mapped on EMSA and the process of each level was described. Monitor, Analyse, Planning and Execute steps in the proposed model respectively act like Perception, Comprehension, Projection and Resolution in EMSA. For simulating the proposed model by agent in Netlogo environment, a notation was introduced and a simulation program was established.

The effect of memory, learning ability and communication as second effective factors in SA was considered in SA of elements of the BISin simulation environment. Based on the results in equal conditions, a system with more learning ability and less memory is more robust than a system with more memory and less ability of learning. As the results of the simulation were shown, communication and learning are the effective parameters in achieving SA. Increasing SA of the BIS elements causes more robustness of them.

References

[1] Dagli C. H., Preface to Part I, Complex Systems, Procedia Computer Science 12 (2012), p. 11 - 12.

[2] Parvar, H., Fesharaki, M. N, Moshiri, B., Shared Situation Awareness System Architecture for Network Centric Environment Decision Making, Secound Int. Conference on Computer and Network Technology, Bangkok, 2010, p. 372-376.

[3] Parvar, H., Fesharaki, M. N., Moshiri, B., Shared Situation Awareness Architecture (S2A2) for Network Centric Disaster Management (NCDM), International Journal of Computer Science Issues, vol. 9, no. 4, Jul. 2012, pp. 503-508.

[4] Endsley M. R., Designing for situation awareness in complex systems, Proceedings of the second international workshop on symbiosis of humans, artifacts and environment, Kyoto Japan, 2001.

[5] Banirostam, T., Self-Protection Modeling Based on Biological Immune System with Autonomous Computing Approach, Ph.D. Thesis, Computer Department, SRBIAU, Sep, 2011.

[6] Endsley, M.R., and Garland, D. J., (Eds.), Situation Awareness Analysis and Measurment, Mashwah: Lawrence Erlbaum Associates, 2000.

[7] Eisenbarth, S. C. and Flavell, R. A., Innate instruction of adaptive immunity revisited: the Inflammasome, EMBO Molecular Medicine, Feb. 2009, p. 92-98.

[8] Banirostam, T. and Fesharki, M., N., A New Approach for Biological Complex Adaptive System Modeling and Simulation, Life Science Journal, 2012;9(3), pp. 2257-2263.

[9] Sun, F. and Xu, F., Antibody Concentration Based Method for Network Security Situation Awareness, 3rd International Conference on Bioinformatics and Biomedical Engineering (ICBBE 2009), June 2009, pp.1-4.