The SmartStreetSensor footfall dataset produced by Local Data Company (LDC) in partnership with the CDRC. The dataset contains details of passive Wi-Fi signal probing from a sensor network across Great Britain. These data are used as a proxy for estimating footfall at retail locations. The sensors capture signals sent by Wi-Fi enabled devices present in their range. The potentially identifiable information collected on the mobile devices is hashed at sensor level and the data is sent to the central server via an encrypted channel for storage.

These data can be extended not only to detect retail activity but to measure all the activity around the sensors and can be linked to transport, work zones and demographic data, etc., to produce, for example, novel functional areas classifications.

While this Secure dataset contains raw sensor data, unique sensor identifiers, raw packet counts and signal strength, we strongly recommend that potential applicants apply for the Safeguarded version of this dataset instead. The Safeguarded dataset includes counts for each sensor location, aggregated to five minute intervals and can be found the related record link below.

Content

The dataset includes details about the location of the sensors (description as well as latitude, longitude, height, depth) and results from probes such as timestamps, MAC address, Vendor OUI, signal strength, and packets.

As of 25 August 2018, there are 1151 locations identified by address (building number, street, unit postcode). The data includes general information about the technical and physical restrictions of each sensor and their locations, anonymised hashed MAC addresses of detected devices captured by the sensors, signal strength and number of packets of detected devices.

Quality, Representation and Bias

The quality of the data is affected by a series of technical limitations relating to the Wi-Fi acquisition process:

1. The range of the sensor: Since the strength of the signal from a mobile device to the Wi-Fi access point varies depending on a variety of factors, the sensors do not have a standard signal range. In other words, the exact delineation of the signal range is different for each sensor at different times.
2. Probing frequency: The rate at which a mobile device probes for available Wi-Fi access points varies widely based on the manufacturer, operating system, state of the mobile device and the number of access points already known to the device.
3. MAC address collisions: There are few instances (<0.01%) of same MAC address being reported by different mobile devices. This might be due to aggressive MAC randomisation by mobile devices and the hashing procedure being carried out twice to sufficiently anonymise the data.
4. Human error: These devices are installed at retail points and they may be disconnected from the main power from time to time, resulting in missing data for certain intervals.
5. Postprocessing: The process to transform probe request into actual footfall requires a series of assumptions that can potentially lead to over or under counting of people in the results. The methodology followed for the data delivered is explained in the next section.

The ten cities with the highest number of sensors are: Central London (318), Edinburgh (46), Wakefield (34), Manchester (32), Glasgow (30), Nottingham (28), Leeds (27), Kingston upon Thames (25) Gloucester (25) and Brighton (23).

The rest of the 500 locations are distributed over 97 other towns across Great Britain. A third of the locations are in Greater London, particularly in London’s central area, which makes any national aggregated count biased towards this region. The date of installation, number of active sensors at any given time and their location can be supplied on request for a particular area. Before July 2016, the number of sensors was limited (no more than 200) and most of these were located in London. A full list of number of locations for each town is provided, see the variable dictionary below.

The sensor cannot distinguish between a mobile device and any other Wi-Fi enabled device. For example, the controlled data includes printers and routers which may be present within the range of the sensor.