All of the following examples are available in NetSim GUI. Navigate to Example > VANETs > Throughput, delay and collisions with SCH and CCH time division. Within Throughput, delay and collisions with SCH and CCH time division users will see four folders. Each folder comprises of simulation samples for Parts 1, 2, 3 and 4 as explained below.
Background#
Dedicated short range communication (DSRC) which uses two channels: Service channel SCH and Control channel (CCH). Each synchronization interval SI is split as follows
Figure 4‑11: We see the time divisions in DSRC. Each synchronization period consists of 1 CCH, 1 SCH and 1 guard interval. While the sync period is generally equal to 100 ms. NetSim allows users to modify the CCH and SCH interval, and in turn the total Sync period.
All devices switch between SCH and CCH and the alternation is based on the time divisions. NetSim allows the user to configure values of CCH interval, SCH interval and Guard interval. The default channels used in NetSim are SCH 171 (5.855 GHz) and CCH 178 (5.890 GHz)
Multiple nodes access the medium using 802.11p protocol. IEEE 802.11p PHY operates in the 5.9 GHz band with a channel bandwidth of 10 MHz 802.11p is an adaptation of the IEEE 802.11a standard used in Wi-Fi systems.
Simulation scenario#
Figure 4‑12: Illustration of the VANET scenario under study. The network comprises of 4 vehicles and 1 roadside unit. Each vehicle transmits two applications: (i) a BSM broadcast application that is sent to all other devices (vehicles plus RSU) within range and (ii) a CBR application transmitted to the RSU
The scenario comprises of four vehicles, V1, V2, V3 and V4 communication to the RSU, R1 and to one another. As explained in Figure 4‑11, each vehicle sends unicast CBR traffic to the RSU and broadcast BSM (safety messages) to one another. Recall that per DSRC functioning, BSM is sent on the CCH while CBR is sent on the SCH.
Simulation parameters and results#
Part 1: Throughput#
The BSM application is configured with packet size of 20B and inter-packet arrival time of 320 μs, while the CBR application is configured with packet size of 1460B and inter-packet arrival time of 5840 μs.
Application | Application Type | Gen. Rate (Mbps) | CCH 20 ms SCH 80 ms | CCH 25 ms SCH 75 ms | CCH 30 ms SCH 70 ms | CCH 50 ms SCH 50 ms |
---|---|---|---|---|---|---|
Throughput (Mbps) | Throughput (Mbps) | Throughput (Mbps) | Throughput (Mbps) | |||
BSM 1 | Broadcast | 0.5 | 0.090 | 0.112 | 0.134 | 0.226 |
BSM 2 | Broadcast | 0.5 | 0.099 | 0.124 | 0.149 | 0.248 |
BSM 3 | Broadcast | 0.5 | 0.105 | 0.131 | 0.158 | 0.264 |
BSM 4 | Broadcast | 0.5 | 0.108 | 0.136 | 0.164 | 0.275 |
Sum Throughput (Mbps) | 0.402 | 0. 504 | 0.605 | 1.014 | ||
Sum Throughput * (SCH+CCH)/CCH | 2.011 | 2.015 | 2.017 | 2.027 | ||
CBR 1 | Unicast | 2 | 1.020 | 1.241 | 1.146 | 0.456 |
CBR 2 | Unicast | 2 | 0.695 | 0.667 | 0.903 | 0.868 |
CBR 3 | Unicast | 2 | 1.379 | 0.972 | 0.594 | 0.612 |
CBR 4 | Unicast | 2 | 1.808 | 1.626 | 1.553 | 1.109 |
Sum Throughput (Mbps) | 4.902 | 4.507 | 4.195 | 3.045 | ||
Sum Throughput * (SCH+CCH)/SCH | 6.127 | 6.009 | 5.993 | 6.090 |
Table 4‑1: We see that the as the CCH interval increases, BSM application has higher throughput rate. Similarly, as the SCH Interval decreases there is decrease in throughput rate.
Observations#
-
BSM in sent on CCH; CBR is sent on SCH. Increasing the fraction of time for CCH increases BSM throughput. Increasing the fraction of time for SCH increases CBR throughput.
-
As expected, Sum throughput divided by SCH fraction is equal for all cases. Similarly, Sum throughput divided by CCH fraction is equal in all cases. This verifies the working of time division between CCH and SCH.
Part 2: Delay#
When analyzing delay, we change the generation rate such that it is below the saturation capacity of the network. If this were not so, then queuing delay would blow-up at (and post) saturation.
Application | Application Type | Gen. Rate (Mbps) | CCH 20 ms SCH 80 ms | CCH 25 ms SCH 75 ms | CCH 30 ms SCH 70 ms | CCH 50 ms SCH 50 ms |
---|---|---|---|---|---|---|
Delay (Micro sec) | Delay (Micro sec) | Delay (Micro sec) | Delay (Micro sec) | |||
BSM 1 | Broadcast | 0.025 | 38774.178 | 33711.174 | 29184.264 | 15302.952 |
BSM 2 | Broadcast | 0.025 | 39085.500 | 33999.099 | 29580.397 | 15452.973 |
BSM 3 | Broadcast | 0.025 | 38871.482 | 34065.658 | 29612.045 | 15592.555 |
BSM 4 | Broadcast | 0.025 | 38971.652 | 34244.557 | 29722.204 | 15687.207 |
Average Delay | 38925.703 | 34005.122 | 29524.728 | 15508.922 | ||
CBR 1 | Unicast | 1 | 970820.956 | 623360.782 | 1070605.026 | 3231622.594 |
CBR 2 | Unicast | 1 | 554478.563 | 1711244.232 | 938543.697 | 3167638.761 |
CBR 3 | Unicast | 1 | 660493.934 | 1021442.963 | 1682596.057 | 3767856.679 |
CBR 4 | Unicast | 1 | 827215.500 | 694044.413 | 1435345.051 | 4040215.389 |
Average Delay | 753252.238 | 1012523.097 | 1281772.458 | 3551833.356 |
Table 4‑2: We see that as the CCH interval increases, the delay for BSM application decreases. Similarly, as the SCH interval decreases the delay for CBR application increases.
Observations#
-
CCH Delay has 3 components (a) waiting time where the packet is waiting for the SCH to complete (b) Medium access time and (c) Transmission time
-
The mathematical analysis of delay is complex. It involves two evaluating two difficult components (i) CCH packet waiting time while SCH packet is served and vice versa, and (ii) medium access time. We leave the mathematical analysis to interested researchers, and restrict ourselves to stating that the trends are as expected i.e., increasing CCH time (and reducing SCH time) reduces the CCH delay (and increases SCH delay)
Part 3: Collision count with increasing generation rate#
The scenario layout remains the same, however we change the application settings. In this example we only have the 4 BSM applications. There are no CBR applications. The application generation rates are mentioned in Row 1 (shaded grey).
Application | Application Type | Gen Rate 0.005 Mbps | Gen Rate 0.010 Mbps | Gen Rate 0.015 Mbps | Gen Rate 0.020 Mbps |
---|---|---|---|---|---|
Collision Count | Collision Count | Collision Count | Collision Count | ||
BSM 1 | Broadcast | 1148 | 2686 | 4347 | 6157 |
BSM 2 | Broadcast | 826 | 1897 | 3049 | 4337 |
BSM 3 | Broadcast | 617 | 1298 | 2223 | 3014 |
BSM 4 | Broadcast | 474 | 1034 | 1670 | 2370 |
Total collisions | 3065 | 6915 | 11289 | 15878 | |
Total pkts transmitted | 28934 | 57818 | 86720 | 115578 | |
Collision Probability | 0.106 | 0.120 | 0.130 | 0.137 |
Table 4‑3: Comparison of Collision count of BSM applications with changing generation rate
Observations#
-
Saturation throughput is about 0.25 Mbps per app or 1 Mbps total. Note the generation rate is below the saturation capacity of the network
-
We see collision probability increases as generation rate increases
-
To the best of our knowledge the mathematical modelling of collisions with non-saturated queues is an open problem. The Bianchi model exists for predicting collision counts with saturated queues, subject to certain conditions.
Part 4: Collisions count with increasing number of nodes#
This scenario has 10 vehicles in a line on a road. The vehicles transmit power Pt = 20 dBm, Carrier sense threshold CSth= − 85 dBm, and we assumed log distance pathloss with η = 2.5. The received power between nodes with maximum separation, d = 100, is
Pr = 20 − 47.88 − 10 × 2.5 × log (100)= − 77.88 dBm
Since Pr > CSth all nodes can hear one another which means that they are all within CS Range.
Results#
Number of Tx nodes | Collision Count | Pkts transmitted | Collision Probability |
---|---|---|---|
1 | 0 | 24786 | 0.000 |
2 | 5297 | 69515 | 0.076 |
3 | 15080 | 127398 | 0.118 |
4 | 28723 | 179751 | 0.160 |
5 | 45665 | 244391 | 0.187 |
6 | 66506 | 316725 | 0.210 |
7 | 89357 | 396539 | 0.225 |
8 | 127606 | 490200 | 0.260 |
9 | 166575 | 586842 | 0.284 |
10 | 216332 | 694672 | 0.311 |
Table 4‑4: Collision probability comparison with change in number of transmitting nodes
The Collision probability is the ratio between Collision count to total number of packets transmitted
$$Collision\ probability = \ \frac{Collision\ count}{packets\ transmitted}$$
Figure 4‑13: Collision probability vs. number of transmitting nodes
Observations#
-
We see collision count increasing with number of transmitting nodes
-
This can be compared against the Bianchi analytical model