Network Simulator Cell

The built-in Eclipse MOSAIC Cell Simulator enables the applications to use cellular network communication. The simulation of cellular communication in Eclipse MOSAIC consists of two parts:

  1. The Cellular Simulator itself and
  2. The applications that can communicate over cellular networks in the Application Simulator

These changes are done in a generic way, making the cellular simulator exchangeable. Users interested in a different kind of simulation of cellular communication may use other simulators and develop ambassadors connecting them to Eclipse MOSAIC.

The Cellular Simulator in the current state consists of three main modules:

  1. UplinkModule
  2. GeocasterModule
  3. DownlinkModule

The Geocaster module simulates a mandatory component for ITS communication. It is inspired by the several architectures from research projects as simTD or CONVERGE to enable ITS use cases over cellular networks. It mainly takes over the task of an addressing and routing component with geographic knowledge to support geo-addressing. However, it also supports regular topological addressing. The Uplink and DownlinkModule are responsible for the transmission simulation. They account for the aspects of transmission delays, packet losses and available data rates. In this context,Uplink and Downlink always refer to the direction towards respectively from the Geocaster. For instance, a transmission from an Internet-based server towards a vehicle would include an Uplink between the server and the Geocaster and a Downlink between the Geocaster and the vehicle. While the Uplink direction only allows point-to-point communication (Unicast), the Downlink supports point-to-point (Unicast) as well as point-to-multipoint (Multicast).

Installation

This simulator does not need to be installed. It is delivered as part of the Eclipse MOSAIC-installation package.

CellAmbassador folder structure

The Eclipse MOSAIC Cell simulator can be configured via three distinct configuration files, which can be found within the scenarios' folder structure:

└─ <scenario_name>
   └─ cell
      ├─ cell_config.json ................. Cell ambassador configuration file
      ├─ network.json ..................... Network configuration file
      └─ regions.json ..................... Regions configuration file

The network and regions configuration files are referenced in the cellular ambassador configuration file.

Read the detailed documentation of the Network and Regions Configuration.

Configuration

We provide a cellular configuration file in the example scenarios of Tiergarten and Barnim. Please note that in the default configuration of this scenario the Cellular Simulator is deactivated. To activate the cellular simulator just enable the cell federate in the scenario_config.json:

"federates": {
    ...
    "cell": true,
    ...
}

The central configuration for the cellular simulator in the file <scenarioName>/cell/cell_config.json could look like this:

{
    "networkConfigurationFile": "network.json ",
    "regionConfigurationFile": "regions.json "
}

Read the detailed documentations of:

The configuration for the global network in the cellular simulator in the file <scenarioName>/cell/network.json could look like this:

{
    "globalNetwork": {
        "uplink": {
            "delay": {
                "type": "ConstantDelay",
                "delay": "100 ms"
            },
            "transmission": {
                "lossProbability": 0.0,
                "maxRetries": 2
            },
            "capacity": 28000000
        },
        "downlink": {
            "unicast": {
                "delay": {
                    "type": "ConstantDelay",
                    "delay": "50 ms"
                },
                "transmission": {
                    "lossProbability": 0.0,
                    "maxRetries": 2
                }
            },
            "multicast": {
                "delay": {
                    "type": "ConstantDelay",
                    "delay": "100 ms"
                },
                "transmission": {
                    "lossProbability": 0.0
                },
                "usableCapacity": 0.6
            },
            "capacity": 42200000
        }
    }
}

The network.json also allows setting the fields defaultDownlinkCapacity and defaultUplinkCapacity, which are used if you don’t specify any bitrates in your CellConfigurations inside your applications. These values are set to 100 Gigabit by default, so if you aren’t interested in capacity/bandwidth observations you can probably just leave them as is.

Server configuration

MOSAIC supports a type of simulation units, called servers. Traffic Management Centers are a specialised type of server. What differentiates servers from other units is, that they are treated as having no geographical location and instead are located “inside the internet.”. This concept omits prior shortcoming, where one would have to configure RSU’s as servers.
Servers are also configured in the network.json as follows:

{
    "defaultDownlinkCapacity": "10 Gbps",
    "defaultUplinkCapacity": "10 Gbps",
    "globalNetwork": {
           [...]
    },
    "servers": {
        {
            "id": "TestServer",
            "uplink": {
                "delay": {
                    "type": "ConstantDelay",
                    "delay": "200 ms"
                },
                "transmission": {
                    "lossProbability": 0.5,
                    "maxRetries": 2
                }
            },
            "downlink": {
                "unicast": {
                    "delay": {
                        "type": "SimpleRandomDelay",
                        "steps": 5,
                        "minDelay": "100 ms",
                        "maxDelay": "200 ms"
                    },
                    "transmission": {
                        "lossProbability": 0.1,
                        "maxRetries": 2
                    }
                }
            }
        },
        {
            [...]
        }
    }
}

The fields defaultDownlinkCapacity, defaultUplinkCapacity are used if you don’t define any bitrates for your cell configurations their default values are 100 Gigabit, so you usually don’t need to set them.

Servers get their configuration for the delay and retransmission models from the network.json and the configuration for the capacity-model has to be made from within an application, by activating the CellModule and properly setting the bit rates. This could look as follows:

        getOs().getCellModule().enable(
                new CellModuleConfiguration()
                        .maxUlBitrate(10000)
                        .maxDlBitrate(10000)
        );

        // Alternatively using default values from network.json
        getOs().getCellModule().enable();

For the CellAmbassador to be able to recognize servers, the group-field in the mapping must match the id-field in the network.json, this also allows for multiple servers to utilize the same configuration. A mapping_config.json using the configuration from above could look as follows:

{
    ...,
    "tmcs": [
        {
            "group": "TestServer",
            "applications": [ "TestApplication" ],
            "inductionLoops": [ ],
            "laneAreaDetectors": [ ]
        }
    ],    
    "servers": [
        {
            "group": "TestServer",
            "applications": [ "TestApplication" ]
        }
    ],
    ...
}

Note that all bandwidths are given in bit per second and all delays in nanoseconds, unless explicitly defined differently. Also, every json configuration goes through a minifier, so comments are allowed in the configuration files. An example would be the comment before the globalNetwork option.

Delay regions

Additionally, the user has the option to define regions with individual delays. This can be used to simulate areas with bad reception, crowded areas etc.

The regions should be stored in <scenarioName>/cell/regions.json. An example definition for a single region called Ernst-Reuter-Platz could look like this:

{
    "regions":[
        {
            "id": "Ernst-Reuter-Platz",
            "area": {
                "nw": { "lon":13.3249, "lat":52.5131 },
                "se": { "lon":13.3273, "lat":52.5125 }
            },
            "uplink": {
                "delay": {
                    "type":         "SimpleRandomDelay",
                    "steps":        4,
                    "minDelay":     "50 ms",
                    "maxDelay":     "200 ms"
                },
                 "transmission": {
                     "lossProbability": 0.8,
                     "maxRetries": 2
                },
                "capacity":         28000000
            },
            "downlink": {
                "unicast": {
                    "delay": {
                        "type":         "SimpleRandomDelay",
                        "steps":        3,
                        "minDelay":     "100 ms",
                        "maxDelay":     "200 ms"
                    },
                    "transmission": {
                        "maxRetries": 2
                    }
                },
                "multicast": {
                    "delay": {
                        "type":         "SimpleRandomDelay",
                        "steps":        3,
                        "minDelay":     "120 ms",
                        "maxDelay":     "220 ms"
                    },
                    "transmission": {
                        "lossProbability": 0.8
                    },
                    "usableCapacity":   0.6
                },
                "capacity":             42200000
            }
        }
    ]
}

Note that nw represents the upper-left (north-west) point of the rectangle and se the lower-right (southeast). For further information about the possible options, please refer to the Eclipse MOSAIC API documentation.

The actual configuration of the uplink and the downlink modules for each region exhibits the identical format as configuration of the globalNetwork in the network.json.

When no regions are found, or if a node (a vehicle) is not within a specified region, the globalNetwork defined in the network.json-File will be used as the delay model.

Transmission simulation

One of the most important feature of the cellular simulator is an estimation of the delay experienced through the transport over the cellular network.

The cellular simulator offers various modes to estimate the delay of the transmissions. The type of estimation is specified with by delayType for the uplink and downlink for each region. You may also refer to the Delay-Chapter.

  • delay.type = ’ConstantDelay’: The message is transmitted with the latency being exactly equal to delay.
  • delay.type = ’SimpleRandomDelay’: The latency can assume different (randomly generated and uniformly distributed) values between minDelay and maxDelay. The number of different values is determined by steps.
  • delay.type = ’GammaRandomDelay’: A gamma distribution is used to estimate the latency, with $ \alpha $ = 2 and $ \beta $= 2. The minimal delay minDelay is added to the result. The curve is fitted so that the maximum likelihood for the delay is exactly equal to expDelay.
  • delay.type = ’GammaSpeedDelay’: This mode closely resembles the GammaRandomDelay. Additionally, a penalty for the velocity with which the node is moving is calculated. This penalty is then added to the original delay. The GammaRandomDelay and the GammaSpeedDelay are derived from a measurement campaign during a research project at the DCAITI.

The two different modes for the downlink are unicast and multicast which are configured separately. Multicast aims to simulate the features of Multimedia Broadcast Multicast Service (MBMS). The main difference in terms of the transmission for unicast and multicast is the handling of undeliverable messages. For unicast, the options lossProbability and maxRetries are used. Pr is short for packet retransmit and denotes the probability for a failed delivery and a subsequent retransmit. The maximum number of retries for the retransmission is configured through the maxRetries-option. The probability of a successful retransmit is recalculated on every try.

In case of multicast the lossProbability is used as packet loss rate. The value is factored into the delay calculation. In contrast to the unicast, just one transmission attempt is made for multicast.

Operation

Beside the transmission simulation, the Addressing and Routing is the other important aspect of the Cellular Simulator. This task is enabled by the Geocaster.

The Geocaster evaluates the message headers for cellular messages, which are created by the communicating applications in the Application Simulator.

It supports the following addressing and casting schemes.

CellTopocast is the normal unicast, where the Geocaster simply resolves the single receiver via theIPResolver. Hence, the CellTopocast directly routes the message further. Currently, Topocast doesn’t allow broadcast or anycast addresses, but any transmission protocols (tcp, udp).

CellGeoUnicast addresses every node in the destination area individually. In this way it takes a geographic address and results in a loop to generate multiple unicasts.

CellGeoBroadcast, which is basically MBMS, uses one broadcast to all nodes in the destined regions.The MBMS uses the different transmission mode of multicast in the downlink. CellGeoUnicast as well as CellGeoBroadcast require broadcast, but don’t allow tcp (as ack for broadcasts is denied).