You are on page 1of 9

Transportation Research Part C 19 (2011) 673–681

Contents lists available at ScienceDirect

Transportation Research Part C
journal homepage: www.elsevier.com/locate/trc

Automatic system for detecting driver use of mobile phones
José Manuel Rodríguez-Ascariz, Luciano Boquete ⇑, Joaquín Cantos, Sergio Ortega
Department of Electronics, University of Alcalá, 28871 Alcalá de Henares, Madrid, Spain

a r t i c l e

i n f o

a b s t r a c t
This paper presents the preliminary results for an automatic electronic system for detecting driver use of mobile phones for voice communications. An electronic circuit (Radio Frequency harvesting) captures the power generated by mobile-phone use and two antennae located inside the vehicle and a signal analysis algorithm are used to identify when the driver is using a mobile phone. Practical tests have been conducted in which the event is detected in the worst operating conditions. The system could be used in research work into automatic detection of mobile-phone use (instead of using autotest or direct human inspection), in automobile insurance as part of a ‘‘Pay-As-You-Drive’’ system, or even in implementation of automated safety devices inside the vehicle. Ó 2010 Elsevier Ltd. All rights reserved.

Article history: Received 2 September 2010 Received in revised form 3 December 2010 Accepted 6 December 2010

Keywords: Driver distraction Mobile phone Signal detection Radio Frequency harvesting

1. Introduction Many studies have shown that driver use of mobile phones increases driving risk (Walsh et al., 2008; Charlton, 2009; Strayer and Drews, 2007; Consiglio et al., 2003; McEvoy et al., 2005). This risk also extends to pedestrians (Loeb and Clarke, 2009; Nasar et al., 2008). For example, it is estimated that mobile-phone use for one hour a month increases accident risk by 400–900% (Violanti, 1998). Other studies show that a high percentage of accidents among youngsters are due to mobilephone use (Neyensa and Boyle, 2007). There is also evidence to show that the risk of a traffic accident while using a mobile phone is equivalent to driving with a blood-alcohol level at the legal limit (Strayer et al., 2006; Redelmeier and Tibshirani, 2001). The increased accident risk is due to the fact that drivers using the phone are distracted from their main task, resulting in slower reaction times to external events. Studies of driver performance when using mobile phones (both hand-held and hands-free units) show that reaction times to events and stimuli are slower (Caird et al., 2008; Al-Darrab et al., 2009). This risk increases in direct proportion to the duration of the call and is also higher during night driving. Although the laws of many countries penalize hand-held mobile-phone use while driving, a significant proportion of drivers continue to do so (McEvoy et al., 2006; Taylor et al., 2003; Gras et al., 2007). In the short term, at least, driver phone use looks likely to increase, so more effective countermeasures are needed (Mccartt et al., 2006). One of these countermeasures could be to prevent the phones from working while a car is being driven (McEvoy et al., 2005). Mobile-phone use can be detected by the following means: questionnaires (self reporting) (Pöysti et al., 2005), sometimes after an accident in the hospital (McEvoy et al., 2005; Bener et al., 2006); simulators (Strayer and Drews, 2007; Drews et al., 2008; Törnros and Bolling, 2006); direct observation by the research team (Taylor et al., 2003); special night vision equipment (Vivoda et al., 2008); driving of vehicles on a closed circuit (Treffner and Barrett, 2004), or other non-automatic methods. To the best of our knowledge, an automatic mobile-phone-use detection system has never been used before.

⇑ Corresponding author. Tel.: +34 918856572; fax: +34 918856591.
E-mail addresses: jmra@depeca.uah.es (J.M. Rodríguez-Ascariz), luciano@depeca.uah.es (L. Boquete), jjcantos@gmail.com (J. Cantos), sergioortegarecuero@hotmail.com (S. Ortega). 0968-090X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.trc.2010.12.002

2. The incident power on the antenna was monitored using an AR power meter (model PM2002) connected to an AR directional coupler (model DC7144A). the power captured is minimal..e..674 J. To maximize the performance of the rectifier. / Transportation Research Part C 19 (2011) 673–681 This paper presents an electronic system for automatic detection of mobile-phone use by drivers. i. 2005). After any of the former solutions the signal is rectified. a combination of antenna and rectifier (Akkermans et al.5 m above the floor level. Other authors use active logarithmic detectors (Hudec et al. In this case. 2 shows a practical example of the power captured in the open air at different distances from a mobile phone while a call is being made. Rodríguez-Ascariz et al. This voltage is digitized at a resolution of 12 bits by an internal ADC on the microcontroller. This process is known as RF Harvesting or RF Energy Scavenging. Fig. A microstrip line is used to achieve impedance matching at 50 X. it is known as a ‘‘rectenna’’.1. manufactured by EAD) that covers the 900/1800 frequency bands and has a 1/4 Wave Element. among other factors. The DC voltage at the output of the rectenna circuit was measured using a Fluke multi-meter (model 289).M. UMTS) and the voltage is amplified by a voltage multiplier (Yan et al. It comprises a commercially available antenna (Mini-GSM model. The rectified DC voltage is stored in a large capacitor and is digitized by an analogue–digital converter (ADC) for subsequent storage and processing using a microcontroller. lowthreshold Schottky diodes (BAT62) are used. 1. which are optimized to work in the GHz frequencies. In its simplest version. 2. When a call is established. on the distance of the phone from the antenna and the relative orientation between antenna and phone.. . The mobile phone and the rectenna for this experiment were placed in a regular university classroom. The voltage obtained with this system depends. which is stored in a condenser (c4). The biggest problem for the system is posed when phones are used by all passengers except the driver. Preliminary experiment and architecture definition The measurements shown in Table 1 were taken to study the circuit’s performance in the GSM (900 MHz) and UMTS (1900 MHz) frequency bands. and these could theoretically be implemented in a car.. Fig. 2. peak gain of 0 dBi and linear polarization. The system for detecting driver use of mobile phones could be integrated into the Pay-As-You-Drive concept as a means of assessing the risk assumed by the driver and adjusting the insurance premium to suit. The electric field was generated inside an ETS semi-anechoic chamber (SpaceSaver RFD-F100) containing an isotropic probe (FL7006) and the detector circuit. A Greinacher multiplier (Curty et al. The antenna is tuned on the frequency bands of interest (GSM. The RF signal was generated by a Rohde & Schwarz unit (SM300 model) then amplified (AR model 50S1G4A) and transmitted inside the semi-anechoic chamber using a BiLog emission antenna (Schaffner–Teseq. Schematic of the rectenna. 2005). 1. Materials and methods Detecting mobile-phone use in a given physical space involves using a power-capturing circuit tuned to the frequency of interest. Commercial systems do exist for detecting mobile-phone use indoors. The detector circuit and the isotropic probe were used to measure the electrical field. this power varies inversely with the distance between the telephone and the power-capturing antenna. 2005) producing a direct current (DC) from RF-electromagnetic energy. The transmitter output power has been selected to meet the GSM standard.. Fig. 2007) is used to rectify the RF signal and obtain a direct current. model CBL6143A). When the telephone is not being used. The trouble is that these commercial systems are not able to discriminate mobile-phone use by a passenger instead of the actual driver. keeping both antennas in vertical position (vertical polarization). 1 shows the electrical circuit implemented. The distance between those devices is shown in Fig. Another application of this system consists of detection of this event and conditional activation of certain vehicle safety measures due to the increased accident risk. a strong electromagnetic field would be created inside the car and the detection system would need to be able to discern that it is not the driver who is using the mobile phone.

34 1800 MHz The values recorded indicate that the rectenna’s response is proportional to the intensity of the electrical field received by the antenna and that it is similar on the two frequency bands studied.17 8.75 12.82 23. the internal structure of the vehicle and its metal chassis interfere with the theoretical ideal distribution behaviour of the mobile-system power (energy proportional to the proximity of the phone). therefore. When the driver adjusts the height of the headrest.29 16.J. varying not only the placement of the antennae but also their orientation. the antenna system is also adjusted. Suitably placed inside the vehicle.84 1. 3. this set-up facilitates more trustworthy discrimination of driver use by appropriate analysis of the captured signals.M. / Transportation Research Part C 19 (2011) 673–681 675 Fig. .04 2.35 1. Frequency 900 MHz Output power tx antenna (dBm) 36 33 30 27 36 33 30 27 E (V/m) 21. The algorithm’s output is transmitted to a personal computer for recording and further study. The mobile phone used was a Sony-Ericsson K800i subscribed to mobile phone operator Movistar.62 4.38 2. Rodríguez-Ascariz et al. 4).98 1. each with its own power-capturing circuit) therefore has to be used (Fig. An integrated 5-V DC battery is used. Power captured at different open-air distances.8 10. A two-antenna system (right and left. Table 1 Rectenna output (VDC) and electrical field (E (V/m)) versus transmitter output power at various frequencies.8 7. 3). 2. A microcontroller carries out analogue–digital conversion and executes the voltage analysis algorithm. Nonetheless. The best detection performance was found to be obtained by placing one antenna each side of the driver’s headrest (Fig. trying to focus the zone of greatest sensitivity on the area around the driver’s head. Two practical experiments were carried out to verify system operation. This involved taking several samples.04 14. Results The system was tested in a 2009 VW Passat on the roads of the outer campus of the University of Alcala. as the power captured is not sufficient to power the microcontroller and its auxiliary circuits. The first involved an assessment of the signals captured when using a single mobile phone inside the car and the second analysed system performance when the phone was used by all vehicle occupants except the driver.88 3. An initial study was carried out to ascertain the best sitting of the system’s two antennae and also to determine signal processing.72 VDC (V) 4.

Diagram of RF harvesting blocks. 5. Results obtained from driver position. Fig. Antenna placement in the car. / Transportation Research Part C 19 (2011) 673–681 Fig. .M.676 J. Fig. 3. 4. Rodríguez-Ascariz et al.

The green line. / Transportation Research Part C 19 (2011) 673–681 677 Fig. first by the driver (Fig. The nearest base station is situated 2. Once the vehicle was moving. shows the signal captured from the antenna fixed on the right-hand side of the headrest and the black line shows the signal captured from the left-hand antenna. a call was made to the car’s mobile phone and answered by the corresponding person. 7. Experiment 1 evaluated the performance of the system when a single mobile phone (Nokia N73. 7). . Figs.J.1. 5). 3. a further breakdown was made depending on which ear the volunteer raised the phone to. 6. then by the rear-seat passenger behind the driver (Fig.2 km far from the mobile phone. the phone was raised to the right and left ears. The vehicle was driven in a circuit closed off to other traffic at a speed of 32 ± 3 km/h (measured with a GPS. Results obtained from rear-seat passenger behind the driver. heights (172 ± 8 cm) and weights (64 ± 11 kg). The call was maintained for 25 s with each recipient speaking for about half the time. The left graph therefore corresponds to the left ear and the right shows the signals captured when the user raised the phone to his or her right ear. 8) and by the front-seat passenger (Fig. 5–9 show the arithmetic mean of the 16 practical tests carried out in each of the five positions in which the occupants could sit inside the vehicle. 9). by the rear-seat passenger behind the front-seat passenger (Fig.M. Within each position. by the central rearseat passenger (Fig. Fig. moreover. model EM-408) and the driver was accompanied by at least the front-seat passenger. Experiment 1 Both experiments were carried out by eight volunteers (four women and four men) of different ages (23 ± 3). 6). UMTS band) was used inside the vehicle. In each of these experiments. Rodríguez-Ascariz et al. Results obtained from central rear-seat passenger.

When the phone was used by other vehicle occupants. As can be seen from the above graphs. a low-pass filter was programmed to eliminate voltage peaks and smooth out the input signal. The system output activates an LED (light-emitting diode) if the above condition is met. however.M. a minimum call limit of 5 s can be set to avoid false activation of the system. 10 show the mean value of the signals captured in the 16 tests carried out. to avoid transitory situations that might trigger false alarms. For example. . 3. 8. The measurements show that mobile-phone use by the driver produces a voltage (4 V) twice as high as use by each of the other vehicle occupants (2 V). 16 tests were conducted in which the vehicle was occupied by 5 people (driver and four passengers) and calls were made simultaneously to the four passengers’ mobile phones but not to the driver’s. / Transportation Research Part C 19 (2011) 673–681 Fig. At the same time. To ascertain how the system performs in this situation.0 V (Fig. The system considers that the driver is using the mobile phone when either of the two signal processing channels is activated at a high level. the voltage levels captured were usually below 2 V. in which case the signal reached a level of 4 V in the antenna closest to the mobile phone and up to 0. distance to base station. barring a small peak at the start of the call when the mobile phone was held to the right ear. A series of preliminary tests were performed to detect inappropriate mobile-phone use inside the vehicle.5 V were detected and unsuccessful when they fell below 3. This is difficult to achieve due to the large number of factors that influence measurement – vehicle chassis. 11). Discussion This paper presents a preliminary system designed to detect when the driver of an automobile is using a mobile phone. The graphs in Fig. it was deemed fitting to implement a hysteresis window in the algorithm to avoid oscillations that might destabilize the system. Results obtained from rear-seat passenger behind the front-seat passenger. This factor makes it possible to identify the spatial position in which mobile-phone use occurs (associated with each of the occupants within the vehicle). etc. model of phone. Experiment 2 The situation most likely to trigger a false alarm is when all vehicle occupants use the phone except the driver. Furthermore. After analysing the graphs with the voltage readings under the different circumstances.2.5 V. Factors considered during implementation are captured voltage levels and temporal characteristics.5 V in the antenna on the other side of the headrest. Rodríguez-Ascariz et al. In Experiment 1. 5.678 J. Detection algorithm The signals captured under real-world operating conditions make it possible to set-up a suitable algorithm in the microcontroller for detecting mobile-phone use. use of the phone by one or more vehicle occupants. Detection was therefore considered to be successful when signals above 3. The passengers raised the phone to both their left and right ears consecutively. This difference enables us to tell when the phone is being used by the vehicle driver. the strongest signal was captured when the phone was used by the vehicle driver (Fig. 5). The low-pass filter used by the algorithm cuts out the peak to pre-empt false alarms. The signals captured were always below 2. mobile-phone users established voice communication non-simultaneously in various locations within the vehicle. 4. the algorithm output is passed to a laptop PC for result analysis.

Rodríguez-Ascariz et al. Fig. Use of four mobile phones inside the vehicle. 10. 9. 11.M. Fig.J. Detection algorithm. . / Transportation Research Part C 19 (2011) 673–681 679 Fig. Results obtained from front-seat passenger.

2008.M.P. IEEE Antennas and Wireless Propagation Letters 4. J. This system may be used to carry out research into mobile-phone use by drivers. 347–355.P. T. 69–75. 187–190. Scialfa. Journal of Safety Research 40. Berg. A... Z. This paper has not considered use of a hands-free mobile phone.R. This paper has not considered use of text messaging (SMS). Visser.. S. 2007.. W. Clarke.. Cell phones and driving: review of research. the proposed system recorded a voltage of 2. J. D. Effect of cellular telephone conversations and other potential interference on reaction time in a braking responses. 2003.. Mobile telephones. TSI2007-61970-C01. M. Acknowledgements The authors would like to express their gratitude to the Spanish Ministry of Education for the support provided through the ‘‘Implementing a Pay-As-You-Drive service using a hardware/software platform’’ Project. Nevertheless. Pechac. M. N. 2007. Driscoll. 1282–1293. P.. McCartt. Accident Analysis & Prevention 40.. P. Haworth. 185–189. Haigney. P. it would be difficult to differentiate between use by the driver and use by the front-seat passenger.J.A.N. during which time mobile communications are not established and no RF data transmissions are made. as part of a Pay-As-You-Drive system. 2005. Current Directions in Psychological Science 16. M.J. 2005. British Medical Journal 331. I.D. 140–142.C. D. M..K... 128–131. 2006. Conclusions This paper presents a preliminary low-cost. 206–212. Woodward. Mobile telephone use among Melbourne drivers: a preventable exposure to injury risk. D. 381–391. Witte.. Boyle.. (Eds. Medical Journal of Australia 185. M. 1–5. 284–290. Hudec. 459–465. M. F. D.. Medical Journal of Australia 179. Springer.A. / Transportation Research Part C 19 (2011) 673–681 Experiment 2 reflects the least favourable situation likely to occur in the vehicle interior. C... P. 47–51. Drews. 2009. Rodríguez-Ascariz et al. Joehl. Bratiman.. The effect of mobile phone use on driving style and driving skills. Bennett. McEvoy.. S. Taylor.. Applied 14. W. Crouch. 2009. F. This situation is highly problematic. Summala.R.A.A. as greatest risk of accident exists when the driver is reading or writing a message.. Redelmeier. R. Nasar. C.. Stevenson. 2006. Al-Darrab.. Strayer. J.. 2001. or as a vehicle safety system. McEvoy.. Drews.. M. The cell phone effect on pedestrian fatalities.. L. 2005... but not the driver) use a mobile phone simultaneously for voice communication. M. 2235–2239.. IEEE Transactions on Microwave Theory and Techniques 53. as in these cases the phone is located relatively far away from the driver’s head area (where the system’s antennae are located).. 2007. Accident Analysis & Prevention 40.A. In this situation. D. 2006. An experimental study on the effect of mobile phone conversation on drivers’ reaction time in braking response... S.... Passenger and cell phone conversations in simulated driving. Palamara. Carter. M. G. A meta-analysis of the effects of cell phones on driver performance. Caird. H.T.. Cell-phone-induced driver distraction. Hecht. Microwave system for the detection and localization of mobile phones in large buildings. 1581–1582. Steel.. 6.N. Doodeman... D..R.. Accident Analysis & Prevention 39. Cercarelli.. P. Font-Mayolas. Charlton. Transportation Research Part E: Logistics 45. small-size system that does not interfere with the vehicle’s other electronic systems. 2007.680 J. Journal of Experimental Psychology. M. Drews. K. 630–634. P. Phone use and crashes while driving: a representative survey of drivers in two Australian states. Dehollain.J. Strayer. 495–500. R. Consiglio.E. Sullman. D..A.. S. It has been proven to detect inappropriate mobile-phone use by the vehicle driver..L. Cunill. namely when 4 out of the 5 occupants (four passengers. Polivka. Work is currently underway on a new test protocol. Loeb.A.P.. References Akkermans. D. 2006. namely when the front-seat passenger uses a mobile phone for voice communication. Bener.. Strayer. Accident Analysis & Prevention 35.A. L.P. such as DCS (Digital Cellular System – 1710–1785 MHz uplink) or DECT (Digital European Cordless Telecommunications – 1880–1900 MHz uplink) by changing the antenna. L. Khan. Factors influencing the use of cellular (mobile) phone during driving and hazards while using it. it has considered the least favourable situation as regards proximity to the antennae. non-invasive. distracted attention.. MMS or e-mail. Lajunen. T. The proposed system minimizes the computational load on the microcontroller....J. Mccartt. Accident Analysis & Prevention 41. Hellinga. Accident Analysis & Prevention 37. Stevenson.I.J. Masupathi. Garewal.. C.A. 2009. Woodward. Zkan. F. Neyensa. R.. Willness. P. as the distance between the mobile phone and the antennae is less than when the hands-free phone is used. Rajalin.G.M. Driving while conversing: cell phones that distract and passengers who react. 2003. Declercq. Canadian Medical Association Journal 164. Traffic Injury Prevention 7. Analytical models for low-power rectenna design.. C. J. W. Design and Optimization of Passive UHF RFID Systems. A comparison of the cell phone driver and the drunk driver.A. Ishrat..5 V. Accident Analysis & Prevention 39... which is not high enough to identify mobile-phone use erroneously as originating from the driver. M. making it possible to implement additional algorithms on the same hardware platform in the future. The limitations encountered to date will be studied in further tests currently being developed to complement the system described in this paper.T.. A. International Journal of Crashworthiness 11..L.L. Human Factors 48... Curty.). M. D. 160–173. The preliminary system proposed in this paper may be adapted to various operating frequencies. M. The resulting trials will study voice communication situations in various vehicle models (as chassis geometry has been revealed to be a significant factor) using different models of mobile phone at varying distances from the base station.M... Aymerich.. M. 2008. Gras.M. Wener... S.G.. Tibshirani. Car phones and car crashes: some popular misconceptions. Ref. H. van Beurden. 2008. 89–106. M. 2005. D. Pöysti. A... Role of mobile phones in motor vehicle crashes resulting in hospital attendance: a case-crossover study. . and pedestrian safety. Mobile phone use while driving in as sample of Spanish university workers. Planes. Moreover. S. The effect of distractions on the crash types of teenage drivers. 392–400.

M. 1998. P. 64–66. 2006. J. A. H. pp. J.M.. de Vreede.P. Eby.J. S. Cellular phone use while driving at night. 298–306. In: Proceedings of the SAFE 2005 Conference. 2004. K.. Rodríguez-Ascariz et al.M. B. Macías Montero. Louis... J. .N.. Kostyniuk... J. 519–524.N.M. Transportation Research Part F 7.. Mobile phone use – effects of conversation on mental workload and driving speed in rural and urban environments. Accident Analysis & Prevention 40. Watson. Traffic Injury Prevention 9. D. Yan. R.C. Treffner. Barrett. Hands-free mobile phone speech while driving degrades coordination and control. White.. Accident Analysis & Prevention 30.K. 2008. J. St.. L..P.. R. A.. Akhnoukh. 2008.. Transportation Research Part F 9. Dialling and driving: factors influencing intentions to use a mobile phone while driving. Cellular phones and fatal traffic collisions.G. Violanti. Burghartz. 37–41.M. / Transportation Research Part C 19 (2011) 673–681 681 Törnros.J. Walsh.W. 1893–1900.. 2005. Hyde. Vivoda. L. Bolling. Utrecht. The Netherlands. 229–246.M.... An integration scheme for RF power harvesting.