Miralles (2002: 225-226) comenta también que el dibujo sirve como indicador y denuncia del

The main metrics for VANET performance used in this work are • the certificate (chain) emission rate,

• the number of packets lost due to cryptographic packet loss, and • channel load in terms of channel busy ratio (CHBR).

Both certificate emission rate as well as cryptographic packet loss are measured within the im- plementation of the VANET protocol stack. The certificate emission rate can relate to pure PSC

or AAC emission or the transmission of a certificate chain, which is in ETSI ITS the combined sending of PSC and AAC.

In contrast, the channel load is measured by equipping nodes within the network simulator with additional wireless devices at the same location as the devices conducting VANET commu- nication. Such extra devices do never transmit, but only probe the wireless channel at each time stamp of the discrete event simulation environment (see also Section 3.3) to determine whether the channel is busy or not. Parameters of such extra receivers are identical to the ones used for ordinary VANET communication.

The caused channel load by an individual node (partly) depends on the average number of certificate inclusions in the security envelope per second. Attacks or cross influence from other VANET functionality may change the frequency of certificate emission. For such cases the relation between emission frequency in case of a present influence and an uninfluenced system is taken into regard.

For all metrics the standard deviation of obtained measurement results is determined and given in corresponding illustrations in later chapters of this work. Thereby, reliability and sig- nificance of the obtained results are illustrated. This also addresses criticism from [92] on the way of presentation of evaluation results in large parts of the VANET related literature.

The metric used in [135], which is called cooperative awareness, is not used within this work. The core reason is that this metric does not consider the data update rate at receivers. I.e,, it is only looked at how many nodes in a dedicated part of a node’s environment use a PSC being known to the ego node, but it is not considered whether the ego node receives any message from these nodes at all. This is especially a problem in urban environments. In such scenarios, two nodes can be very close (e.g., closer than 100 m), but no message exchange is possible due to shadowing from a building. Such nodes have a negative influence on the cooperative awareness as defined in [133, 135], although there is never any message exchange between both nodes. Therefore, no cryptographic packet loss occurs and the presence of such nodes should not be counted as a negative performance criteria of the PSC distribution strategy.

The next section introduces the traffic scenarios, which are used for evaluations throughout this thesis.

3.2

Traffic Scenarios

An overview of popular traffic scenarios in prior work is given in Section 2.4.3. The following road network topologies are considered within this work.

• freeway scenario: three lanes in each direction and 6 km length (see Figure 3.4), e.g., used in [106, 190]

• rural road: a straight road with one lane in each direction, 6 km length and extra vehicles joining the road about in the center of the scenario, like suggested in [204] (see Figure 3.1). The joining roads have only one lane with traffic going towards the main road.

• urban grid: represents Munich Schwanthalerh¨ohe (see Figure 3.3), as exported from Open Street Map on 17th July 2014 [241]. Urban grid scenarios are used, e.g., in [288, 304].

• urban roundabout: represents the quite large roundabout to be found in Munich Maxvorstadt (see Figure 3.2), as exported from Open Street Map on 17th July 2014 [241]. Usage of this kind of scenario is suggested in [93].

Traffic on the road topologies freeway and rural road is determined by defining deterministic traffic flows within SUMO. In contrast, traffic for the much more complex topologies of urban grid and roundabout is created using the SUMO random trip generator1.

Figure 3.1: Road topology of the rural road scenario.

Figure 3.2 shows the roundabout scenario (SUMO screen shot). The yellow triangles repre- sent vehicles on the road. Random trips start and end at the edge of each road leading towards the roundabout, and each start position is connected with each possible end position of a trip.

The urban grid scenario is displayed in Figure 3.3. Comparison to the roundabout scenario (see Figure 3.2) shows that this scenario is more affected by shadowing. Both scenarios share low to medium mobility of nodes, due to applied speed limits according to urban environments. In regard to velocity profiles a maximum velocity of 50 km_h is set in the urban roundabout and the urban grid scenarios for all vehicles. The rural road scenario uses speed limits of 100

km

h on the rural road itself and 50 km

h on the roads being connected to the main road at about the center of the scenario. In the freeway scenario different traffic flows with individual maximum velocities are used on the dedicated lanes following the recommendations in [106]. The most right lane is preferred by vehicles using a maximum velocity of 80 km_h , vehicles on the middle lane yield 110 km_h , and the most left lane is preferred by vehicles going up to 130 km_h .

Parameters for traffic flows are derived from traffic densities. In doing so, the traffic density is varied from 16 to 45 _kilometervehicles [146]. The freeway scenarios include the ones recommended in [106]. To obtain results for VANETs realized within the given traffic scenarios, the simulation environment discussed in the next section is used.

Figure 3.2: Road topology of the urban roundabout scenario.