Skip to content
Permalink
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
170 lines (126 sloc) 4.2 KB

MASSim Protocol Documentation

The MASSim protocol defines sequences of JSON messages which are exchanged between the agents and the MASSim server. Agents communicate with the server using standard TCP sockets.

Sequence

An agent may initiate communication by sending an AUTH-REQUEST message, containing its login credentials.

The server tries to authenticate the agent and answers with an AUTH-RESPONSE message.

If the result was positive, the server now sends various messages over the socket depending on its state:

  • a SIM-START message, whenever a new simulation starts
  • a REQUEST-ACTION message right before each simulation step
  • a SIM-END message after each simulation
  • a BYE message when all matches have been run, just before the server closes all sockets

An agent should respond with an ACTION message to all REQUEST-ACTION messages it receives (within the time limit).

Reconnection

If an agent loses the connection to the server, it may reconnect using the standard AUTH-REQUEST message. Auhtentication proceeds as before. If authentication was successful and the agent reconnects into a running simulation, the SIM-START message is sent again. If it coincides with a new simulation step, the order of SIM-START and REQUEST-ACTION messages is not guaranteed.

Message formats

Each message is terminated by a separate 0 byte. The server buffers everything up to the 0 byte and tries to parse a JSON string from that.

Each message is a JSON object following the base format

{
   "type": "some-type",
   "content": {...}
}

where value of type denotes the type of the message. The JSON object under the content key depends on the message type.

AUTH-REQUEST

  • Who? - Agents
  • Why? - Initiate communication with the server.
{
  "type": "auth-request",
  "content": {
    "user": "agentA1",
    "pw": "1"
  }
}
  • user: username of the agent that is configured in the server
  • pw: the agent's password to authenticate with

AUTH-RESPONSE

  • Who? - Server
  • Why? - Tell agent the result of authentication after an AUTH-REQUEST message.
{
  "type": "auth-response",
  "content": {"result": "ok"}
}
  • result: the result of the authentication; either "ok" or "fail"

SIM-START

  • Who? - Server
  • Why? - Indicate that a simulation has started or is currently running.
{
  "type": "sim-start",
  "content": {
    "time": 1489514146201,
    "percept": {
      ...
    }
  }
}
  • time: when the message was created
  • percept: static simulation info

The contents of the percept object depend on the scenario (see Percepts section of scenario.md).

REQUEST-ACTION

  • Who? - Server
  • Why? - Indicate a new simulation step, convey simulation state and request an action.
{
  "type": "request-action",
  "content": {
    "id": 2,
    "time": 1556636930397,
    "deadline": 1556636934400,
    "step": 27,
    "percept": {
      ...
    }
  }
}
  • deadline: the time by which the server expects an ACTION message
  • id: the action-id; this id must be used in the ACTION message so that it can be associated with the correct request (which prevents older actions from being executed if they arrive too late)
  • step: the current simulation step
  • percept: the current simulation state

The contents of the percept object depend on the scenario (see Percepts section of scenario.md).

ACTION

  • Who? - Agent
  • Why? - Respond to REQUEST-ACTION message
{
  "type": "action",
  "content": {
    "id": 2,
    "type": "move",
    "p": ["n"]
  }
}
  • id: the action-id that was received in the REQUEST-ACTION message
  • type: the type of action to use
  • p: an array of parameters for the action

SIM-END

  • Who? - Server
  • Why? - Indicate that a simulation ended and give results.
{
  "type": "sim-end",
  "content": {
    "score": 9001,
    "ranking": 1,
    "time": 1556638423323
   }
}
  • ranking: the team's rank at the end of the simulation
  • score: the team's final score

BYE

  • Who? - Server
  • Why? - Indicate that all simulations have finished and the sockets will be closed.
{
  "type": "bye",
  "content": {}
}
You can’t perform that action at this time.