This distribution of AgentScape is a modified version of AgentScape version 0.9.0beta2. It was made specifically for a demonstration that was presented during the demo track at the AAMAS-2007 conference: the conference on Autonomous Agents and Multi-Agent Systems. As such, no support is given for this distribution. Feel free to experiment with it, but for support please use the official distribution of AgentScape that can be downloaded from the AgentScape website. It is likely that some features of this demo distribution may end up in the official distribution in the near future.

For more information on the demo see also http://www.agentscape.org/aamas07/demo.html

This demo runs on Linux, MacOS and Windows. The requirements to run this demonstration, are as follows:

• Sun JRE 1.5

Look at the terminal window to monitor any errors. If all goes well, the simulation demo will appear on screen.

When the demo starts, it randomly starts a number of producer services and consumer agents. The blue boxes represent the individuals: consumer agents are on the left, and the producer services are on the right. You can add more of these using the both various 'add' buttons. An individual instance can be removed be clicking the respective close button:

For both consumers and producers some settings can be set that will be valid during the next simulation. Consumers have a fixed maximum energy demand per hour, and they can behave in a static way (always demand the same amount) or in a dynamic way (where demand varies over the simulated day – though never exceeding the maximum demand). The behaviour of the agent can be toggled between dynamic and static by clicking on the green agent icon to toggle between dynamic agents () and static agents ().

Producers have some more settings that can be applied. First, there are 3 available energy producer types:

• diesel generators – these will always produce the same amount

• solar panels – these have a maximum production capacity, but will produce less energy during the evening and night than during the day

• wind generators – these start out with a fixed maximum capacity, but the production can fluctuate when there is a lack of wind.

Each producer has a PPU field that determines how much a producer charges an agent for a single unit of energy. The consumers can prefer to buy energy from the cheapest provider if they want to.

Once all the consumers and producers are set up, press the 'run' button to simulate a single day for the producers and consumers. During a day, you cannot manipulate the consumers and producers. Once the simulation starts, the clock starts running and for each hour agents will negotiate with the producers for energy.

For each hour, a line is drawn between a consumer and a producer. A black line means that a single producer was chosen to supply the energy. Blue lines means that offers from multiple providers have been aggregated to fulfill a single request. As time goes by, a graph will be displayed for both the consumers and producers. For consumers, a blue bar indicates the amount of energy that was obtained, and red means that a certain amount could not be delivered. Similarly for producers, blue indicates the amount of energy that was sold, and red indicates an overproduction. The graphs also reflect the difference in behavior between various strategies.

The negotiation settings area allows for some global settings to be tested. Most interestingly, the resource aggregation can be disabled to inspect the difference in behavior of the entire simulation. Here it can also be set which providers and consumers are preferred to be handled first. Consumers probably want to select the cheapest offer, but when pricing does not matter, it can also be used to prefer the provider with the most/least amount free. Not all of these strategies have a large impact if resource aggregation is enabled.

Depending on the strategy, it should be possible for the producers to adapt their pricing and/or production dynamically. A producer could notice that is has a constant overproduction, and as a result it can lower the price or the production, in order to prevent overproduction. This has not been implemented in the current version of the simulation, but doing so can always be done using the existing framework.

To learn more about mediated service access, have a look at the publications at www.iids.org

AgentScape details

This distribution contains the source release of AgentScape, pre-compiled with Java 1.5. Documentation about running AgentScape and agents can be found in the agentscape/doc directories. Running the demo only starts a platform for the purpose of the demo (the simulation agent). Once the demo agent finishes, the entire platform will be shut down. You can run a normal agentscape platform as well from this distribution, and even alter and recompile AgentScape if you so desire (ant and javac are needed).

The AgentScape tutorial can be found in the agentscape/doc directory. This includes information on how to write agents in agentscape, how to start and setup an agentscape platform and explanations of the example agents.

Negotiation Framework

Contact

This distribution was produced by the IIDS research group, department of Computer Science, VU University Amsterdam.

For more information about (other) IIDS research and AgentScape see:
www.iids.org
www.agentscape.org