The adoption of new 5G standards in Belgium is a hot topic. Many supporters cite braking distances for self-driving cars as a good reason to kick 4G to the curb, as 5G ensures significantly lower latency than 4G, as well as guaranteed quality of service. But is latency really putting drivers at risk? We did the test!
 
The idea behind the concept of connected cars is that each vehicle interacts with other vehicles nearby to share their positions. Doing so allows a car to notify passengers in the event of an emergency stop, for example. But the longer it takes for one car to communicate with other cars, the greater the necessary braking distance becomes. Some 5G supporters use this fact to claim that latency is causing unsafe conditions for drivers. We were intrigued by this claim and wanted to put it to the test.
 

Does the 4G network increase braking distances?

So, we designed an experiment to explore just how much braking distances increase due to latency in the current 4G network. Armed with a smartphone, a ByteBlower Wireless Endpoint and a public ByteBlower port, myself and my colleague hopped in the car. As we drove, another colleague launched a test protocol to measure network latency.

The results of our latency-measuring road test 

After five minutes of driving, we drove back to the Excentis lab and had a look at our measurements. The results reveal some interesting things: after about two minutes on the road, we saw a large increase in latency probably caused by mast switching. We also saw large differences in latency across the entire period, and high jitter. But what does this mean for braking distances? Does our 4G network support safe driving with connected cars?

Figure 1: Example of a ByteBlower latency measurement over a 4G connection .

Using our measurements, we put together a case. Driving at a speed of 120 kph, we’d be traveling at 33.33 meters per second. In the best-case scenario, the latency at this speed would be 14 milliseconds – doubled to be 28 milliseconds between connected cars. Using these figures, we estimate that the braking distance would be extended by 0.93 meters.

The average latency in our test was 27.56 milliseconds, doubling the braking distance increase to 1.8 meters. However, the maximum latency measured during our brief test exceeded 126 milliseconds, increasing the braking distance by 8.4 meters! 

Update 21/3/2019:
The result above is measured on an best effort 4G connection. Using an SPS link with controlled latency on 4G would be a better comparison.

Are the braking distance increases significant?

5G promises latencies of between 1 and 2 milliseconds between the car and the transmitting station, adding up to 2 to 4 milliseconds between two cars, or an extra braking distance up to 13 centimeters. For comparison, the reaction time of humans in traffic is 1 second – equal to a traveled distance of 33.3 meters in our experiment.

So, we could say that latency-caused delays in detecting an emergency stop is negligible, because human reaction time is so much slower. Connected car braking performance using the current 4G network is a huge improvement to human reaction time. Still, it must be said that 5G networks will promise an even safer future!

Figure 2: Braking distance at 120km/h caused by human reaction vs. network latency. 

More information

ByteBlower can help you test network performance. Read how ByteBlower makes testing easy or get in touch for more information.

 

Add new comment