17 October 2017
The answer is…It depends on your needs? Before I explain this I will run through some little-known truths about each scanning method and the learnings I have obtained from research and development done on Addinsight hardware.
Addinsight - Traffic Intelligence System is a data analytics solution tailored to the needs of a road agency. We are not a private company and we do not sell hardware. We are a road authority (South Australian Department of Planning, Transport and Infrastructure) and the software was developed in-house to meet the needs of a road authority. We encourage hardware vendors to develop hardware to communicate with our system, so that Addinsight users have greater choice. The current list of certified hardware suppliers can be found here.
Firstly to Bluetooth. This methodology captures Bluetooth devices such as hands-free kits, car stereos and GPS devices that are in discovery mode. You will notice that I did not mention smart phones. Here is the first little-known truth. Even with Bluetooth turned on, smartphones are rarely in discovery mode, making them invisible to Bluetooth scanners. They are only in discovery mode when you are on the Bluetooth settings page trying to pair to another device. Once two Bluetooth devices have gone through the pairing process (e.g. car stereo and phone), they can discover each other without needing discovery mode.
This is the reason why Bluetooth scanning only gets about 10-15% of the total traffic and not 50% or more. This is also why Bluetooth scanners are not effective in pedestrian only environments; the primary device you are trying to track is a smartphone and the scanners can’t see them. Not being able to detect smartphones does sometimes work in its favour. In a mixed environment, your traffic data will get far less noise from pedestrians, cyclists and buses full of people with smartphones.
Contrary to some published reports, it doesn’t take 10 seconds to detect a Bluetooth device. Multiple devices can be found within the first few seconds of a scan. Just try a scan on any smartphone and results will appear almost immediately.
The detection of discoverable Bluetooth devices tends to be very reliable. By this I mean that if a device passes near a Bluetooth scanner, even for a couple of seconds, there is a high chance it will be detected. Anecdotally, there is about a 10% chance of a device not being detected at each site, so more often than not you will detect that same device as they travel along a corridor.
Now to Wi-Fi. Unlike Bluetooth, Wi-Fi will detect smartphones. You immediately think…winner! But that is not the whole story. Here is the second little-known truth. Wi-Fi scanners use passive scanning, so it is up to the phone to decide if it wants to look for an access point or not.
When you analyse the data collected by a Wi-Fi scanner in an office environment the detection rate is huge. But when you look at the raw data you notice a pattern. Most devices are discoverable for less than a second and then they disappear for 60 seconds or more. In most cases it is a visible pattern at regular 60 second intervals. As long as the device is in range for close to 60 seconds, there is a very high chance of detection, which is why it works so well for pedestrians.
In a signalised road environment, Wi-Fi site detection rates are about the same as Bluetooth. But when you start to pair the data between sites, Wi-Fi falls behind. That is because the devices detected at site A may be a completely different set of devices to site B. In high speed environments, devices are in range for only a few seconds and there is a very high chance of a device not being detected at each site. Your population of trackable devices is bigger but you will be finding them randomly and not reliably.
Wi-Fi also has a couple of extra problems. Firstly, Software generated MAC addresses. When in sleep mode, some devices do not use their hardware MAC address to find access points, so you may be detecting a random MAC address that will not appear again down the road. The other problem is that Wi-Fi will detect phones on pedestrians, bikes and in vehicles, so in a busy city location you will get a lot of noise and you won’t know what you are tracking. A single bus full of passengers could generate a lot data.
As you can see, there is no right answer that covers all situations. Here are my suggestions. Use Bluetooth for the reliable detection of road traffic, particularly in high-speed environments. It’s terrible for non-vehicle detection. Definitely use Wi-Fi for this. Wi-Fi can be used in low-speed traffic environments, but just don’t expect reliable detections for all traffic situations. Where possible use both.
Addinsight - Traffic Intelligence System will be at the ITS World Congress in Montreal at the end of October and we would love for you to come and have a chat. We will be located at the ITS Australia stand with SAGE Automation. If you would like to know more information about the system please go to www.addinsight.com.au or follow Addinsight on LinkedIn to keep up with the latest updates.
I intend to post more articles in the lead up to the ITS World Congress. As a preview, we are hoping to be able to provide some fresh results of a new hardware device that will be released soon by SAGE Automation. Initial field trials suggest the hardware may achieve 300% more travel time pairings than standard Classic Bluetooth scanners.