Detecting the use of Intentionally Transmitting Personal Electronic Devices Onboard Commercial Aircraft Randy Woods Jay J. Ely Dr. Linda Vahala Old Dominion University NASA Langley Old Dominion University Norfolk, Virginia USA Hampton, Virginia USA Norfolk, Virginia USA [email protected] [email protected] [email protected] Abstract detectors carried by the flight crew, and second an installed The need to detect unauthorized usage of intentionally system. While a portable system might provide the ability transmitting portable electronic devices (PEDs) onboard to single out a particular offender, the system will be commercial aircraft is growing, while still allowing ineffective during critical stages of flight when the crew passengers to use selected unintentionally transmitting must remain seated. Furthermore, information from devices, such as laptop computers and CD players during Alitalia Airlines shows that they were only able to find the non-critical stages of flight. The following paper presents offender to within half of the aircraft 22% of the time and, an installed system for detecting PEDs over multiple identify a seat row 8% of the time with a portable system frequency bands. Additionally, the advantages of a fixed [1]. verses mobile system are discussed. While data is In contrast, a permanently installed system will provide presented to cover the frequency range of 20 MHz to 6.5 continuous uniform coverage of the entire passenger cabin GHz, special attention was given to the Cellular/PCS allowing the attendant to transverse the airplane visually bands as well as Bluetooth and the FRS radio bands. searching for the violator when an alarm is received. This Measurement data from both the semi-anechoic and will prove to be more valuable in implementation due to reverberation chambers are then analyzed and correlated the ability to monitor the entire passenger cabin including with data collected onboard a commercial aircraft to lavatories. determine the dominant mode of coupling inside the Leaky coaxial cable is used in many applications such passenger cabin of the aircraft versus distance from the as tunnels and buildings to provide a uniform coverage that source. As a final check of system feasibility, several would not otherwise be possible in complex PEDs’ transmission signatures were recorded and electromagnetic environments. While the cables do act as compared with the expected levels. antennas, there are undesirable characteristics such as high Introduction insertion loss and larger cable loss per unit length than Cellular telephones and other intentionally radiating conventional coaxial cable. However, leaky coaxial cables devices have the potential to cause interference onboard do present several design aspects that are well suited for the application of detecting PEDs onboard aircraft, such as a aircraft by coupling power into sensitive receivers due to fairly uniform coverage across all common commercial the high allowance for spurious radiated emissions into bands and unobtrusive system design. The system design aviation radio frequency bands (up to –13 dBm)[4]. This coupling of signals can cause interference with critical presented here tries to minimize the undesirable effects by flight systems and therefore, needs to be detected before shorting cable runs to 100 ft and placing the receiver in the causing an anomaly. A second concern is the increased center to minimize cable losses. While taking full advantage of the desirable characteristics such as uniform radiating distance cellular/PCS telephones have while coverage over multiple frequency bands. airborne could cause disturbances with ground based Two types of leaky coaxial cable were tested for this cellular towers by broadcasting to several towers simultaneously. With the possibility of serious disruptions paper. First, Radiax® Cable (Figure 2) provided by to systems and there not being a clear necessity for their Andrews © and second, FlexRad (Figure 3) provided by use, the Federal Communication Commission (FCC) has Times Cable ©. Table 1 compares some of the physical characteristics of the different cables tested. banned cellular phones for use on airplanes [3], and the Federal Aviation Administration (FAA) prohibits all intentionally transmitting devices unless the aircraft operator has determined non-interference [2]. There are two general approaches for detecting the use of intentionally transmitting devices. First, portable 231 ft (70.4m) 2 1 20ft 100 ft Lea1ky0 0C ofatx Cable run Receiver 100 ft Leaky Coax Cable run (6m) 2 Figure 1 Approximate Dimensions of 747 Aircraft Passenger Cabin Inner detection system due to increased cable loss, free space loss Solid Conductor and, receiver complexity. For obvious reasons there will outer need to be slight design modifications depending on the Windows in conductor exact model of aircraft the system is installed in, but the outer electromagnetic characteristics should remain conductor approximately constant due to similar geometry of the different airframes. Approximate dimensions for the main Figure 2 Diagram of Radiax Cable passenger cabin of a 747-400 aircraft are shown in Figure 1. These dimensions allow for an efficient design consisting of a single leaky coaxial cable run along the length of the passenger cabin. A single cable run minimizes weight and installation time, while also reducing the complexity of installation. Due to the electromagnetic characteristics of the passenger cabin, it will be shown that Figure 3 Diagram of FlexRad Cable the physical location of the cable will have little effect on system performance. Using two 100 ft sections of cable, Table 1. Radiax Cable Specifications and placing the receiver in the middle gives the two Andrews Andrews Times extremes shown in Figure 1. RXL4.5- RXL1- Flex-Rad Location 1: Passenger closest to the receiver 1AX 1A 600 (Prevention of false alarms): The desired system response (Cable 1) (Cable 2) (Cable 3) will allow this person to use an unintentional transmitter Diameter 0.865 in 0.30 in 0.52 without triggering an alarm. Due to this requirement the (in) minimum sensitivity of the receiver will be set according to Weight 0.15 0.055 0.09 the specifications given in part 15.109 (Figure 5) of the (lb/ft) FCC regulations for unintentionally transmitting devices. Additionally, the system needs to be dynamic enough such that a received signal of several dBm will not cause adverse effects such as amplifier damage. 55.0 54.0 960 10000 53.0 52.0 51.0 uV at 3m4444567890.....00000 216 960 B45.0 d44.0 43.0 88 216 42.0 41.0 40.0 30 88 39.0 10 100 Frequency (MHz) 1000 10000 Figure 4 Cable Comparisons FCC Limits for Non-intentional Trnasmitters Figure 5 FCC limits for Non-intentional Transmitters System Design The 747 aircraft fuselage was used as a design model Location 2: Passenger farthest from the receiver due to a large number in service and it is representative of a (Detection of unauthorized transmitters): Desired wide body type commercial aircraft. Large aircraft present system response is to detect any intentional transmitter the most difficult design constraints for a distributed PED (Assumed to be ‡ 0 dBm). At location 2, the attenuation 17 0.00 16 0.00 uV) 15 0.00 B d h ( 14 0.00 gt n Stre 13 0.00 d el 12 0.00 Fi 11 0.00 10 0.00 0 0.2 0 .4 0 .6 0 .8 1 D is ta n ce (m ) D irect F ield S treng th R everbe ra nt F ield R esultant F ie ld Figure 6 Field Strength Inside Passenger Cabin of 707 Aircraft from cable loss, and the free space distance between the 8(cid:215) p P E = (cid:215) 5(cid:215) Max Received transmitter and the cable are both at their maximum values, EST l h resulting in the lowest received signal strength at the cable, rx E =Field Strength (V/m) and the largest cable loss, resulting in the lowest received EST (2) power for an intentional transmitter that must be detected. l = Wave Length Estimated Field Strength h = Antenna Efficiency rx To determine the distance where the dominant mode of P =Maximum Power Received Max Received coupling switches from direct coupling to a reverberant Using time averaging, the data in Figure 7 was field, it is necessary to simulate the electromagnetic collected inside the reverberation chamber. This shows the environment of the passenger cabin. This was done by insertion loss inside a reverberant field across frequency comparing the power density inside the reverberation with mechanical stirring. While the power density is not chamber with the average power density collected by Naval shown, Figure 7 does show the uniform insertion loss into Surface Warfare Center (NSWC) [5] inside the passenger the cable. The precise insertion loss is determined using cabin of a 707-720B aircraft (The passenger cabin of the the data collected inside the semi-anechoic chamber where aircraft tested was essentially intact, but stripped of all the electric field is easily calculated. seats). The presence of 200 seats in the hull of an aircraft will reduce the average cavity gain by about 1.26 dB at 100 0 MHz and 0.89 dB at 5.8 GHz [7] therefore, seating -5100 2100 4100 6100 -10 configuration will have little effect on the chamber -15 characteristics. -20 Using Equation 1 [9], power transmitted is converted B d-25 to electric field strength assuming an isotropic radiator and -30 no reverberant effects. Then using the average power -35 density (-52 dBm/cm2) and Equation 2 [8], the estimated -40 reverberant field strength of the aircraft was superimposed -45 Frequency (KHz) on the graph resulting in Figure 6, where the reverberant Times cable 5/8 inch Radiax 1/4 inch Radiax cable field is dominant greater than 0.1 meters from the Figure 7 Reverberation Chamber received Power transmitter. An additional property investigated in the reverberation P = E 2 (cid:215) 4p (cid:215) R 2 ⇒ chamber was the minimum distance from a metal surface PRP 120 (cid:215) p where the leaky coaxial cable is effective as an antenna. (1) 30 (cid:215) P This was determined by measuring the cable loss (Figure 8) E = and insertion loss (Figure 9) at several heights from the R 2 metal floor using a 15-foot segment of cable. P = Equivalent peak radiated power (W) PRP E = Electric field intensity (V/m) R= Distance in meters from source 0 -0.5 -1 -1.5 -2 B d -2.5 -3 -3.5 -4 -4.5 0 1000 2000 3000 4000 5000 6000 Frequency (MHz) 0 cm .32 cm 0.64 cm 1.27 cm 2.5 cm 5 cm Figure 8 Cable Loss Vs. Height From Metal Surface The cable loss is measured by connecting one end of Figure 10 Semi-Anechoic Chamber Insertion Losses the leaky coaxial cable to a spectrum analyzer and the other   ( ) end to a signal generator. The cable loss is then the system V (dBmV)=V dBmV - V 1 (3) Received Incident m Coupling m throughput loss. Insertion loss is measured by terminating field Factor one end of the leaky coaxial cable and radiating a signal to This gives a direct correlation of field strength determine the level received by the spectrum analyzer (dBm V/m) outside the cable to received voltage (dBm V) at through the cable. Concluded from these two tests is that the terminal of the Leaky coaxial cable and is not the cable performs uniformly at distances greater than 2.5 dependent on the distance of the transmitter from the cable. cm from a metal surface. Instead, it is a property of the cable itself. 0 Broadband Frequency Coverage and Null Patterns 0 1000 2000 3000 4000 5000 6000 7000 -10 In order to compare mode stirred reverberation chamber -20 field strength data with airplane measurements data, it -30 needed to be determined if there will be sufficient modal -40 dB-50 stirring provided in the passenger cabin to justify using -60 mechanical stirrers while collecting data in the -70 reverberation chamber. While there are not mechanical -80 stirrers onboard an aircraft, several other sources of modal -90 Frequency (MHz) stirring occur, such as flexing of the aircraft’s hull while in 0.8 m from floor on floor 2.5 cm from floor 5 cm from floor flight, and relative movement between the transmitting Figure 9 Insertion Losses inside Reverberation antenna and the aircraft will provide some stirring. Chamber at Various Heights from Floor To investigate if relative movement between the Once the effects of the reverberation chamber were antenna and the reverberation chamber boundaries recorded the leaky coaxial cable was placed in a semi- provided a corresponding shift in the null pattern received anechoic chamber to determine the insertion loss into the by the spectrum analyzer, the leaky coaxial cable was cable. Using calibration data provided by the placed in the reverberation chamber and the transmitting manufacturer, the antenna’s field strength (dBm V/m) at the antenna was moved around a small area showing the cable was determined for a distance of 1 and 3 meters. development of the nulls for different antenna locations Once the field strength present outside the cable is known, without mechanical stirring. As the antenna was moved, it the insertion loss into the cable in dB can be determined became obvious that the pattern of nulls also shifted. using Equation 3. Figure 10 shows the insertion loss for Additionally, this phenomenon is observed in data various locations along the cable as a function of distance collected inside the semi-anechoic chamber. Where by from the cable termination. collecting data at several locations along the length of the cable, Figure 10 show that the nulls present at the spectrum analyzer are not stationary but are dependent on the location where the signal is coupled into the cable. The combination of the two results provides confidence that passenger movement and airplane flexure will remove nulls received by a detection system as a function of time averaging and therefore justifies the use of mechanical stirring while inside the reverberation chamber to average Link Budget out nulls. The power received is dependent on several factors such as power transmitted by the PED, insertion loss of the 0 -10824 829 834 839 844 849 leaky coaxial cable and, free space loss at the particular B)-20 frequency. From Figure 6 it is seen that the free space d ss (-30 distance only effects field strength at distances of less than o-40 L n -50 0.1m. Figure 13 shows the actual received device levels o erti-60 inside the semi-anechoic chamber. Using these values, ns-70 I Table 2 shows the estimated insertion losses for actual -80 -90 devices in the frequency bands of interest. Frequency (MHz) 0 Radiax insertion loss Limit based on 95% confidence -10 0 500 1000 1500 2000 2500 3000 Figure 11 Cellular Band Null Pattern Corrected for -20 Airplane Losses Power Received (dB) -----7654300000 FRRadSi o - 41.0 CPheollun-le3a r8 . 8 PPChoSn - e4 9.5 -80 -78.0 -90 Bluetoo th device -100 Frequency (MHz) Figure 13 Device Signatures for Frequency Bands of Concern Table 2 Received signal strength for actual PED devices using 5/8” Radiax Cable at 1m Figure 12 Weibull Probability Distribution for Cellular Telephone Band Approximate Power Insertion To ensure system feasibility even if there is not power Frequency Band Received losses sufficient stirring provided onboard the aircraft, Figure 11 transmitted (dBm) (dB) shows a plot of the cellular telephone band without (dBm) mechanical stirring. This shows the deep nulls that develop FRS Radio 27 -41 68 without modal stirring. Additionally, Figure 12 shows a (» 460 MHz) Weibull cumulative distribution function, which was used Cellular 27 -39 66 to determine the 95% confidence level without stirring. (» 830 MHz) With these findings it was determined that even if there is PCS not sufficient stirring provided onboard the aircraft, that the 27 -50 77 (» 1860 MHz) system design would still be sound with a 95% confidence level (shown as the red line in Figure 11). Bluetooth 0 -78 78 (» 2400 MHz) Detector Design While a superhetrodyne receiver would be a good From Table 2, the limiting received power is for a choice due uniform coverage across a wide frequency Bluetooth device. Due to the desire to cover several range, the proposed receiver instead uses multiple filters to frequency bands using the same antenna there will be continuously monitor all frequency bands. This design additional losses associated with a power divider depending decision is based on the requirement of the superhetrodyne on the number of separate bands covered. To cover the receiver requiring a local oscillator. While a receiver frequency bands listed in the table for instance a 1:4 power system could be designed to meet the FCC requirements for splitter will be required. The loss associated with the non-intentional transmitters, it is thought that a passive splitter is an additional 6 dB bringing the true received system will present advantages during certification for power to approximately –84 dBm. This will require a onboard use. The link budge is calculated based on the use preamplifier to boost the signal strength by 30-40 dB, of multiple filter circuits. which will allow the signal to be sent to a power meter with a tangential signal sensitivity (TSS) of approximately –50 dBm. The output of the power meter is a small DC voltage aircraft across all frequency bands of interest. Due to the (» 10 mV). Since the output of the power meter is limitations of the reverberation chambers the data below proportional to the input power the signal can drive an 100 MHz will need to be verified during further testing. instrumentation amplifier with the output of the instrument While the data shows that the environment inside the amplifier connected to a comparator and finally an alarm aircraft is reverberant and, therefore, could be monitored as circuit. efficiently with a single antenna. The uniform field is not consistent throughout the length of the aircraft due to the Noise Analysis many apertures (windows) in the hull and walls dividing To determine the noise floor of the system several the sections. This is overcome with the leaky coaxial assumptions were needed. system described since the signal decays in a predictable 1. The Bluetooth band is used as the limiting case for fashion once coupled onto the cable. Several other factors noise. This is due to the Bluetooth band’s wide are unaccounted for in this analysis that are present in a real frequency band of 2.4 to 2.5 GHz. To cover the entire environment such as passengers and other chamber loads. frequency band, a band pass filter with a bandwidth of These loads will result in an overall lower Q value for the 100MHz will be required. passenger cabin. 2. An amplifier with an assumed noise factor (NF) of 4.0 Due to the space constraints onboard an aircraft the is used. leaky coaxial cable allows for an antenna the size of the 3. The transmission line loss is assumed from the base of cable that would otherwise be used to carry the signal back the leaky coaxial cable to the receiver. This from another system. While the cable does not perform assumption follows from the way insertion loss was particularly well over any one frequency band it was able found to be at the termination of the leaky coaxial to cover a frequency range that required three conventional cable and can therefore be assumed as small (@ 1 dB). antennas to produce the test signal, thus providing a good 4. Due to the large noise factor of the amplifier, the compromise between space required and overall system antenna noise temperature will not have a large effect performance. on the system. To compensate for the antenna noise temperature, the final result was rounded up. References [1] De Donno, Fabio, Capt., Seventh Flight Safety Conference, “Portable To calculate the input noise floor, the following 3 equations Electronic Devices on Aeroplanes”, 2000 from Reference Data for Engineers [10] was used. [2] FAA AC 21, www.faa.gov/avr/afs/acs/ac91211a.pdf T =T +(L(cid:215) F- 1)(cid:215) T ( 4) E A 0 [3] FCC 14CFR15 section 22 T =Effective Noise Temperature of System E http://www.fcc.gov/oet/info/rules/ T =Antenna Noise Temperature [4] FCC CFR 22 section 917, A L=Transmission Line Loss http://www.access.gpo.gov/nara/cfr/waisidx_02/47cfr22_02.html F = Noise Factor of Receiver [5] Hatfield, Michael O. Naval Surface Warfare Center, Dahlgren, “Phase T =290 K II Demonstration Test of the Electromagnetic Reverberation o Characteristics of a Large Transport Aircraft”, NSWCDD/TR-97/84 N =kBT (5) [6] Ladbury, John. “Evaluation of the NASA LARC Mode-Stirred i E dBm =- 198.6+10log(B)+10log(T ) (6) Chamber Facility”, NIST, January 1999 i E k=1.38· 10- 23 [7] Nguyen, Truong. “RF Loading Effects of Aircraft Seats in an B=Bandwidth in Hz Electromagnetic Reverberating Environment”, NASA Langley Research Center Hampton Virginia, T =Effective noise temperature E NASA TM-2000-21062 Using Equation 4, TE is calculated to be approximately [8] RCTA DO-160D Section 20, “Environmental Conditions and Test 450 K. Then using Equations 5 and 6, the noise floor is Procedures for Airborne Equipment”, Prepared by SC-135, July 29, calculated to be -92dBm . Since the received signal for a 1 1997 Bluetooth device is estimated to be at –84 dBm, the signal [9] RCTA DO-233, “Portable Electronic Devices Carried Onboard will appear slightly above the noise floor. Aircraft”, Prepared by SC-177, 20 August 1996 Conclusion [10] Reference Data for Engineers, Eight Edition, SAMS publishing 1993 Based on data collected, leaky coaxial cable is capable of performing as a distributed PED detector onboard an