Bluetooth tracking for surveys

With traffic surveys remaining a key component of the planning and appraisal process, the introduction of Bluetooth tracking – developed primarily as an alternative to temporary automatic number plate recognition ­– has had a real impact on cutting costs

In austere times, it has been vital to reduce the costs of transportation studies by making more use of existing data sets and developing methodologies that offer added value. However, traffic surveys remain a key component of the planning and appraisal process. Clients want more for their money, and  results delivered quicker.

One development which has had a real impact on cutting costs for some types of traffic survey is Bluetooth tracking, developed primarily as an alternative to temporary automatic number plate recognition (ANPR) surveys, which are among the most expensive to deploy.

How does it work?

The fundamental principle of Bluetooth tracking is straightforward – virtually every mobile phone, hands-free kit, in-car telephone preparation, PDA and laptop, and many Sat Nav and other devices, have a unique MAC (Media Access Control) address that, when enabled, can be detected by a Bluetooth recording device. Bluetooth MAC addresses are detected at the points of interest and then matched between sites to provide origin and destination data, and journey time data.

The sophisticated elements of the survey come in both the field set-up and the data processing to derive and match the required data. Specialist field calibration ensures accurate data capture and complex filtering algorithms analyse the captured MAC addresses to ensure that multiple devices detected simultaneously are treated as one. This is essential to eliminate any potential bias from say a person or vehicle with several devices. Typically, capture rates are 30 per cent for traffic and 15 per cent for pedestrians, although this is expected to rise over the years in line with increased Bluetooth usage.

Big brother

Whilst Bluetooth is certainly tracking of people’s movements, motorised or not, the technology is non-intrusive and anonymous to individuals. It’s a methodology developed to understand how people move around, but without knowing who they are.

Unlike ANPR, a MAC address is non-personal. It’s a unique code used for devices to communicate exclusively between each other, and each MAC address is truncated and encrypted at source, adding further security and anonymity. The technology ticks all the data protection boxes.

Deployment and applications

A Bluetooth scanner is typically deployed in an elevated position to capture the desired detection area. The power and range of the scanner can be adjusted, so the detection footprint can range from a highway or reduced to focus on an escalator or shop front. Bluetooth scanners are self-contained and the equipment casing is compact and inconspicuous. As Bluetooth captures a sample of the flow, it is normal to have a supporting background count to factor up the Bluetooth data.

Bluetooth tracking is suitable for a wide range of transportation data collection exercises, including origin-destination studies, journey time surveys, and public transport user surveys.

To date, the need for a reliable alternative to ANPR for the collection of origin-destination data has aroused most interest in Bluetooth technology. There are a number of reasons that Bluetooth has become a recognised, and in some quarters preferred, alternative to ANPR:

• timescale and trend analysis;

• reliability;

• cost;

• data accessibility and speed of reporting;

• health & safety.

The higher specification Bluetooth units report the data back, up to every minute if necessary. This improved granularity of data means that more accurate journey times can be reported in real time.

Significantly, a Bluetooth device may be carried by an individual from start to finish on the same journey, irrespective of transport mode, so Bluetooth technology can bridge the gap between traffic and pedestrian movement, and there are particular benefits for the study of public transport interchanges. Reporting combined journey times also has applications for events and the retail sector.

Aside from origin-destination recording, Bluetooth tracking is ideally suited to journey time data collection, with careful deployment of monitoring devices serving as a viable alternative to either ANPR or floating vehicle/ moving observer methodologies. The massive benefit that is derived from Bluetooth is the ability to collect literally thousands of journey times per link as opposed to the relatively small number journeys that are commonly collected per survey period with the floating vehicle method.

The technology is also gaining a foothold as a means of recording pedestrian movements and routing, particularly in connection with public transport hub and interchange areas, but also for movements around and within stadia, shopping malls, and public open spaces. This data has traditionally been problematic to record, necessitating both extensive camera coverage and a significant post-survey processing exercise. Streetwise has undertaken a number of pedestrian tracking surveys using GIS (Geographical Information Systems) where surveyors track and spatially record a sample of specific routes taken around the study zone, also recording demographic information. This serves as an exact record of movement and activity, and this compliments the Bluetooth data from strategically located detectors. The spatial interrogation of both data sets in GIS then serves as a comprehensive data set for the study of pedestrian movement.

There is also potential for using the technology in tandem with other survey types. A widely recognised limitation to the effectiveness of roadside interview data is the inability to record mainline motorway data. The typical address to this problem is to survey on feeder sites, motorway ramps where possible. However, this is sometimes done ‘blind’ with no real knowledge of what proportion of traffic passing the target site originates from each selected feeder site. A pre-survey using Bluetooth recording would serve as a means by which to identify suitable and valuable survey sites. Given the high cost, directly for the survey and indirectly due to congestion, of roadside interview sites, any scope to reduce the number of sites to ensure that only relevant sites are selected would surely be of interest to the commissioning authority. Furthermore, being able to ‘bolt on’ other detectors, at a specific location such as a car park,  can assist with any specific stated preference aspect of a study. Being able to scale a study in this manner can add great value and partially answers questions normally only achieved by interview or questionnaire.

Suitability and classifications

Of course, inevitably, there are pitfalls with the technology, and it will not be suitable for all studies. The major issue for most survey types is the inability to differentiate between different vehicle (or indeed non-vehicle) classifications. The recording unit and associated software has no way to identify whether a MAC code was from a device being carried in a car, van, goods vehicle, or even pedestrians pocket. It is possible, dependent on the route and availability of suitable mounting locations, to be able to time-synch a video camera with the Bluetooth unit to be able to classify MAC codes as the device is detected. However, the post-processing will lead to significant delay and the methodology will only work with relatively quiet, uncongested routes as the spread of the Bluetooth capture zone make it impossible to definitively identify where on the road the vehicle was picked up. In urban environments the class issue should not be so important as journey times will be relatively similar across classes, and setting time parameters can easily exclude pedestrians from the data set, unless the cordon is very small.

Having deployed Bluetooth on major studies in the UK and Ireland, the choice of equipment location and the setup and calibration process is critical to collect and indeed isolate the data of interest.

Significant benefits

Whilst we recognise that the technology will not address the needs of every study, Bluetooth matching is here to stay as a technology because it offers some significant advantages over alternative survey types.

Firstly, as outlined in the introduction, Bluetooth matching for O-D surveys is usually cheaper per day of data collected than recording the same information using a number plate survey. The fewer number of units, the ease and safety with which they can be deployed, and the streamlined analysis processes ensure that this is a very cost effective method.

Secondly, this technology is usually far more time efficient than ANPR. Less resource on site typically means lead-in times will be shorter, but primarily the time saving is post-survey. For a long time the A in ANPR has been misunderstood as a process that can be conducted instantly by automatic means. This is almost never the case, as footage typically needs to be run through specialist software to achieve a ‘first pass’ match rate of rarely more than 80% (for a cordon study), with the remaining records being manually in-filled during an ntensive process. This is in stark contrast to a data set that can be scheduled to download immediately after, or even during the survey period, and then run for the required outputs. Whilst it would be unusual to happen in practice, there is the potential for no working time loss with data being presented the morning after the survey.

Thirdly, there is the issue of risk of data loss. Any large scale cordon survey conducted by ANPR is at the significant risk of loss of footage. Since the replacement of VHS with digital recording by virtually all survey companies, there has been an increase in system failure. Footage loss of 10% has been typical within some recording systems, and though this has improved as more robust digital recording units have been introduced, loss of footage is nonetheless a major concern for a cordon survey where loss of a single site can invalidate an entire study of scores of cameras. This issue is largely avoided with Bluetooth, firstly because the recording units have less to do than a camera/ DVR system and therefore have less that can go wrong, secondly because the data can be checked live for consistency.

This issue of survey duration is another major benefit that is delivered by this technology. As well as being able to collect information on weekend movements and record habitual behaviour during the survey period, recording over multiple days minimises the risk of abortive surveys caused by extraneous factors such as untypical travel patterns caused by weather or accident/ incident within the study area.

There have been instances in the past where surveys in the five figure cost bracket which had to be entirely repeated at the client’s cost due to events causing an abandonment of the study. With a 7-day Bluetooth survey, this data would simply be set to null and ignored, as other days provide additional and data which effectively eliminate this risk.

Viable and robust

Bluetooth for transport and pedestrian surveys has emerged as a viable and robust methodology, and has come of age in the past couple of years, and it is likely that it will represent an increasing share of the market for diverse survey requirements. In the field of transportation, Bluetooth is a cost effective evaluation tool for pre and post scheme evaluation. For pedestrian and cyclist tracking, the retail sector, and understanding the dynamics of crowd movement, innovation has enabled new doors to be opened.

It’s a learning curve for everyone, so skilled experience in it’s deployment and use is vital to realistically advise on clients needs and expectations, but more importantly, to advise if Bluetooth is the right tool for the job.