Temporally Autonomous Agent Interaction

Temporally Autonomous Agent Interaction

Adam J. Conover (Towson University, USA) and Robert J. Hammell (Towson University, USA)
DOI: 10.4018/978-1-60960-171-3.ch002
OnDemand PDF Download:
List Price: $37.50
10% Discount:-$3.75


This work reflects the results of continuing research into “temporally autonomous” multi-agent interaction. Many traditional approaches to modeling multi-agent systems involve synchronizing all agent activity in simulated environments to a single “universal” clock. In other words, agent behavior is regulated by a global timer where all agents act and interact deterministically in time. However, if the objective of any such simulation is to model the behavior of real-world entities, this discrete timing mechanism yields an artificially constrained representation of actual physical agent interaction. In addition to the behavioral autonomy normally associated with agents, simulated agents must also have temporal autonomy in order to interact realistically. Intercommunication should occur without global coordination or synchronization. To this end, a specialized simulation framework is developed. Several simulations are conducted from which data are gathered and we subsequently demonstrate that manipulation of the timing variable amongst interacting agents affects the emergent behaviors of agent populations.
Chapter Preview

Message Activation Model

In this mode, each agent begins in a random Boolean state conforming to the basic “Conway” life/death (active/inactive) Game of Life rules (Gardner, 1970). As with the threaded model discussed in previous work (Conover, 2008a), the agents behave autonomously within a global mean vivification delay time 978-1-60960-171-3.ch002.m01 of 500ms with delay variances 978-1-60960-171-3.ch002.m02 chosen to produce 978-1-60960-171-3.ch002.m03 ratios 978-1-60960-171-3.ch002.m04 ranging from 0.0 to 2.02. However, instead of agents simply examining their neighborhood at intervals which are independent of the environment, the agents trigger the vivification of their neighbors by sending event messages. To maintain temporal autonomy, each agent periodically queries an internal message queue (once per vivification cycle) for the presence of pending notifications received from other agents. The agent adopts a new state from the statistical mode derived from the queued messages as well as its current state. If an agent is inactive, it cannot become active until it receives a notification from an active neighbor. Only active agents are capable of sending messages to other agents. When any given agent vivificates, it determines the state of its own environment and sends notifications to all neighbors, if it becomes or remains active. An agent will only send one message to each of its neighboring agents once per vivification cycle regardless of how many messages are in the queue. Once the vivification cycle completes (all neighbors have been notified), the sending agent clears its message queue and again awaits new messages from neighboring agents.

The primary focus of this section is an exploration of the average population density of active agents and average age of the agents as a given trial progresses. However, in this section, the number of messages received by each agent between vivifications is considered. A summary of the data gathered in the first set of message based activation trials is shown in Table 1, ordered by978-1-60960-171-3.ch002.m05 Other values include the average population density978-1-60960-171-3.ch002.m06, the population's average age978-1-60960-171-3.ch002.m07, the average number of messages received per agent978-1-60960-171-3.ch002.m08, and the standard deviations978-1-60960-171-3.ch002.m09, 978-1-60960-171-3.ch002.m10, 978-1-60960-171-3.ch002.m11, of data in each sample set grouped by 978-1-60960-171-3.ch002.m12.

Complete Chapter List

Search this Book: