How can cellular towers detect a mobile phone user’s mobility?
Mobile phones communicate with a cellular tower through which they connect to the rest of the world. During voice calls and data transfer, the mobile phones exchange signaling data with cellular towers for service operation. Thus, wireless service providers can infer the instantaneous position of each mobile phone from the cellular network signaling data. Depending on effort and spatial granularity, there are at least three approaches to extract the mobility information of mobile phone users from cellular networks:
Call detail records (CDR): Each record is generated when a mobile phone user performs an activity, such as making a voice call, sending a text or accessing the Internet. In CDR, the cell-ID indicates which cell tower is serving the users. Thanks to the Google Maps Geolocation API, which returns a location and accuracy radius based on information about cell towers that the mobile client can detect, we can map cell-ID to a physical location as the estimation of the mobile phone user’s location. The spatial granularity of this approach depends on the radius of a cell tower ranging from 100 meters to 1000 meters. But this approach does not impose any additional cost because wireless service providers store CDRs in dedicated databases from which they can be easily retrieved.
Cell change update (CCU): In order to be reachable, mobile phones always inform the network whenever they change cells regardless of whether the mobile phone is in active or idle state. This approach may require additional monitoring infrastructure to dump the network link messages. But with this approach, the data samples can extend to the moving subscribers who do not frequently use their cell phones.
Network measurement report (NMR): Mobile phones periodically measure received signal strength (RSS) from surrounding towers and send it back to the associated tower to create a network measurement report (NMR). By analyzing NMR, wireless service providers will not only know the user’s position at a cell level, but also can improve the accuracy of position with triangulation algorithms, such as the time difference of arrival (TDOA). Like CCU, NMR is not as available as CDR.
What is the advantage of cellular data?
Unlike GPS data, the accuracy of positioning a mobile phone by cellular data is so awkward that it is hardly useful for most location-based services applications. However, the cellular data has many advantages:
Low cost: As mentioned, cellular data is readily available to wireless service providers without incurring additional cost on the network infrastructure or mobile devices. In contrast, GPS data comes with a significant energy cost. This cost is inherent because GPS is based on communication with satellites flying in an orbit 11,000 miles above the earth. The distance difference between a satellite and a typical cellular tower directly translates into 117 dB energy consumption at the receiver. According to an in-house test with Accenture employees in the Beijing office, mobile phones with GPS actively running in the background consume energy twice as fast as they normally do.
Sample size: Thanks to proliferation of Apple iPhone and Android, the number of mobile phones has increased exponentially to the extent that almost everyone owns and uses mobile phones. If using the mobile phone as a mobility sensor, the tracked objectives could be as large as the entire human population living on earth. The sample size is simply the same number as the mobile phone subscribers.
Coverage: Like sample size, the coverage is not limited to any geographical area. For major wireless service providers such as AT&T, Verizon in the US or China Mobile in China, their service covers the entire country. With cellular data, we could collect mobility information from practically anywhere in a country.
What can cellular data be used for?
Analyzing cellular data for position information has a variety of applications. Due to privacy concerns, the existing applications typically provide aggregate mobility information and claim to be anonymous. Despite this, the cellular data has proved its value in a wide range of areas. Examples include customer behavior analysis for retail, property, leisure and media, business location selection for retail, and trip matrix analysis for city/transportation planners. Specific examples include:
As more people use mobile phones and cellular networks coverage expands, the value of cellular data will continue to grow. Potential directions may be congestion detection, real-time traffic estimation, geo-based targeted advertising and family/employees tracking.
To what extent the cellular data can be used really depends on two factors: accuracy (i.e., spatial granularity) and speed (i.e., how fast you can track a cell phone). The accuracy is limited by the time resolution in the cellular network. The 2G/3G networks record a timestamp only in milliseconds, which is not good enough for triangulation algorithms. But the situation will change with 4G networks or Long-Term Evolution (LTE) standards. On the other hand, speed is definitely a challenge for big data technology. An easy calculation will tell us the challenge: tracking 1.22 billion mobile phone users in China for real-time position updates is obviously more difficult than Facebook keeping real-time updates for its 800 million users.