Channel fading measurement and statistics of the eCall minimum set of data transmission
2019; Institution of Engineering and Technology; Volume: 14; Issue: 1 Linguagem: Inglês
10.1049/iet-com.2018.6022
ISSN1751-8636
AutoresYunrui Li, John Q. Liu, Jacob C. Brandenburg,
Tópico(s)Advanced Data Compression Techniques
ResumoIET CommunicationsVolume 14, Issue 1 p. 127-134 Research ArticleFree Access Channel fading measurement and statistics of the eCall minimum set of data transmission Yunrui Li, Corresponding Author Yunrui Li yunrui.li@wayne.edu Electrical and Computer Engineering Department, Wayne State University, Detroit, USASearch for more papers by this authorJohn Q. Liu, John Q. Liu Electrical and Computer Engineering Department, Wayne State University, Detroit, USASearch for more papers by this authorJacob C. Brandenburg, Jacob C. Brandenburg General Motors Corporation, 30003 Fisher Brothers Road, Warren, USASearch for more papers by this author Yunrui Li, Corresponding Author Yunrui Li yunrui.li@wayne.edu Electrical and Computer Engineering Department, Wayne State University, Detroit, USASearch for more papers by this authorJohn Q. Liu, John Q. Liu Electrical and Computer Engineering Department, Wayne State University, Detroit, USASearch for more papers by this authorJacob C. Brandenburg, Jacob C. Brandenburg General Motors Corporation, 30003 Fisher Brothers Road, Warren, USASearch for more papers by this author First published: 01 January 2020 https://doi.org/10.1049/iet-com.2018.6022AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract This study presents a testbed for the emergency call (eCall) system satisfying the 3GPP TS 26.267/268/269 standard. Experiments are performed with the in-vehicle system testbed in a laboratory or a car travelling in city, suburb, countryside or freeway. Sufficient data of the emergency communication signal is recorded and processed. Fading statistics of the received signal after power control are found and discussed with cumulative distribution function of the received signal power, level crossing rate and average fade duration. It is found that with probability less or equal to , fading and attenuation can vary from dB for the continuous wave (CW) signal at 500 Hz to dB for the CW signal at 2000 Hz. It is recommended to move the CW signals at 500 and 800 Hz for detection and synchronisation in the 3GPP standard to be at 1500 and 2000 Hz, respectively. This will give 9.5 dB improvement in detection and synchronisation. The longest fade duration observed is 540 ms. The 64 ms synchronisation signal in the current standard is insufficient to handle the fading. It is shown that fading is minimised in the frequency range [1000, 2500] Hz for the 3GPP TS 26.267 system. 1 Introduction Mobile emergency communication systems are needed for human society to handle emergencies such as car accidents, life-threatening medical situations, hurricanes or earthquakes. In 1996, General Motors (GM) launched the OnStar mobile emergency communication services in the USA employing in-band modem to transmit emergency data from vehicle to the call centre. In 1997, BMW started to offer emergency communication services using the short message service for emergency data transmission. Other automotive manufacturers have followed the practice and the automotive emergency call (eCall) has become popular worldwide. With the strong support of the European Union (EU) government, the first international standard in the eCall was released by the third generation partnership project (3GPP) standard group as TS 26.267, 26.268, 26.269 in 2009. On 28 April 2015, the European Parliament approved a regulation which mandates all new type-approved vehicle models in M1 and N1 class vehicles in the EU market to be installed with eCall devices satisfying the 3GPP standard by 31 March 2018 [1]. When an accident happens, the in-vehicle system (IVS) can automatically detect the accident and dial the European unified eCall number 112 and report the location of the vehicle through transmitting the minimum set of data (MSD) to the public safety answering point (PSAP). The expected benefit is that the rescue service can obtain precise accident position data and reach the accident location sooner. The quicker response will save thousands of lives and lessen the severity of injuries in the EU [2, 3]. The eCall is expected to have great social benefits of relieving human suffering after car accidents and reducing healthcare and other related costs [4]. The EU delegated the 3GPP to standardise technical specifications of the eCall data transmission. After careful study, the 3GPP group selected the in-band modem solution operating through the voice channel of 2G/3G cellular network and public switched telephone network (PSTN) for the emergency data transmission [5]. This is the best choice and the right technical approach to minimise delay of emergency data transmission. The digital cellular voice channel is circuit-switched and the signal transmission is real time. The capacity of the digital cellular voice channel is sufficient for emergency data communication [5]. In 2G/3G cellular networks, the voice channel is of higher priority than data channels. The voice channel coverage is larger than the coverage of data channels. Data transmission through the cellular voice channel is very challenging and demands more research to understand the channel. The EN15722:2015 standard defines the emergency data as the MSD which is a packet of 140 bytes. In the eCall system requirements, it is written that 'The MSD should typically be made available to the PSAP within 4 s, measured from the time when end to end connection with the PSAP is established' (Section 4.2, 3GPP TS 26.267, Version 15.0, June 2018 [5]). This requirement has remained unachievable since the publication of the first version of the standard in 2009. The road test results presented by Harmonized eCall European Pilot (HeERO) project showed that the mean of the MSD transmission delay was 13 s and the minimum duration was 8 s, while the success rate of the MSD delivery was 71 for long number [6]. As this system is for emergency communications, the failure in MSD delivery definitely needs to be improved. Fundamental research needs to be performed to understand the channel before the requirement of delay in the standard can be met. A testbed was designed and built by our team to study the performance of the EU eCall system [7]. Extensive experimental work has been completed using the testbed. It was found that large fading still exists in the voice channels of the global system of mobile telecommunications (GSM) after power control. The fading is not negligible. For emergency data transmission, the fading can cause severe performance degradation in signal detection, demodulation and decoding. It must be measured carefully and counted in system performance evaluation for emergency data transmission through digital cellular voice channel for systems like the 3GPP TS 26.267. In [8], authors discussed the eCall system requirements and the selection process of the in-band modem solution by the 3GPP. The in-band modem solution proposed by Qualcomm was selected after comparing three proposals through PC-to-PC computer simulations. Many realistic factors in the digital cellular network were not considered in the selection process, such as echo canceller, noise canceller, handover and effect of transcoding between codecs employed in digital cellular networks and PSTN. In [9], the uncoded bit error rate (BER) and signal-to-distortion power ratio (SDR) were obtained for the eCall modem through the GSM adaptive multi-rate (AMR) vocoder and additive white Gaussian noise (AWGN) channel by simulation. It was shown that the AMR vocoder can cause significant SDR degradation to the 3GPP in-band modem output signal. For example, when the AMR vocoder rate is 4.75 kbps, the SDR at the vocoder output is 4 dB for the robust modulation mode or 3.9 dB for the fast modulation, causing the demodulation error rate to be or , respectively. In [10], field test results in Finland were reported. It also reported that the success rate of the MSD transmission varied significantly at different test sites. The success rate of the MSD varied from 71 to 100% among the seven eCall pilots of the HeERO project. It pointed out that 'there is a significant potential for improvement in the success rate of MSD transmission in real-life operating conditions'. In [11], authors applied the Gilbert-Elliott model to the voice channel and obtained the capacity bounds for in-band modem data transmission through the voice channel. It also proposed an improved detection scheme invoking Weibull and Chi-square distributions. Multipath propagation fading between cellphone antenna and base transceiver station (BTS) radio receiver without power control was studied in [12, 13]. Anti-fading techniques have been employed to decrease the impact of fading [14], which include power control [15, 16], diversity [17], interleaving, channel coding [18] and so on. The received analogue voice signal through the cellular voice channel meets the quality requirements for voice transmission, but it is yet unknown whether fading impacts on the in-band modem solution significantly. Existing technical specification [5] and core standards only describe requirements, architecture and simulation conformance tests on the EU eCall system. The 3GPP eCall conformance tests and previous data modulation schemes treated the digital voice as if it were an AWGN channel. In our experiments, it was found that although the GSM system employs power control for its voice channel, strong fading still exists in the output signal of the GSM voice channel. This fading affects the success rate of MSD transmission. The impact of fading in the GSM voice channel output on the performance of the EU eCall MSD transmission is yet to appear in the literature. In this study, experiments are designed to measure the signal power at the PSAP demodulator input and analyse the statistics of fading in the GSM voice channel after power control. Section 2 describes the system model of the testbed implementing the EU eCall. Section 3 gives measurement apparatus and experimental procedure. Section 4 provides a statistical analysis of the power of the GSM voice signal received by the PSAP. Cumulative distribution function (CDF), level crossing rate (LCR) and average fade duration (AFD) are obtained for the power of the received GSM voice signal after power control. Section 5 concludes the study. 2 System model The block diagram of the EU eCall system is shown in Fig. 1. The electronic control unit (ECU) in the IVS continuously monitors a trigger signal from multiple types of sensors or an emergency button in the vehicle. When an accident occurs, the IVS is triggered automatically by the sensors or manually by pressing the emergency button to dial the eCall number 112. An eCall is established with the highest priority between the IVS and the PSAP call centre through the cellular network and the PSTN. The ECU reads the vehicle identification number, GPS coordinates, time stamp, type of vehicle, number of passengers and other relevant information and constructs the MSD following the standard. The pull mode is configured as the first operation mode. After the eCall has been answered, the PSAP shall send 'START' signal to notify the IVS to send the MSD. Once the IVS detects the 'START' signal, the uplink MSD signal including the synchronisation frame and the MSD data frames, shall immediately be transmitted to the PSAP through the digital cellular voice channel. The synchronisation frame contains a synchronisation preamble and a synchronisation tone of 500 Hz for fast modulation mode or 800 Hz for the robust modulation mode. The microphone in the IVS is muted during the transmission of the MSD to minimise noise and interference. The PSAP receiver detects the signal, performs demodulation and decoding for the MSD. If the decoding is successful, the PSAP sends an acknowledgement (ACK) signal to the IVS. Upon receiving the ACK signal, the IVS switches to talking mode for people in the vehicle to talk with the person at the PSAP for further assistance. Fig. 1Open in figure viewerPowerPoint Block diagram of the up-link channel for real-time emergency data transmission in the EU eCall system The maximum size of the MSD is 140 bytes represented by 1120 binary bits. The binary bit stream of the MSD is used to calculate the cyclic redundancy check (CRC) parity bits. The CRC encoder output 1148 bits CRC appended MSD to a Turbo encoder. The coding rate is 1/3, and the Turbo encoder generates 3456 bits MSD packet with 12 trellis bits. A hybrid automatic repeat request (HARQ) scheme is also applied to the channel coded bits to create eight different redundancy versions with 1380 bits. The modulation scheme employed in the EU eCall is bipolar pulse position modulation [9] (1) where is the pulse shaping function is the floor function, the sampling rate is equal to 8000 Hz and is the number of symbols (2) The spectral expression of is (3) where is the roll-off factor. The 1380-bits HARQ encoded data is modulated into symbols with three bits per symbol. The symbol time is 2 ms in the fast modulation mode or 4 ms in the robust modulation mode [5]. Therefore, the modulation rate is 1500 or 750 bits/s for the encoded binary data stream. The IVS modem output signal is fed into the speech encoder, which is a highly non-linear device and can cause severe distortion to the MSD signal. The speech codec output signal is then modulated by the Gaussian minimum shift keying (GMSK) modulator for transmission through the radio transceiver. There are multiple propagation paths between the IVS antenna and cellular tower antenna. The impulse response of the channel can be expressed as [14] (4) where is the attenuation factor of the kth path of the L paths and is the time delay of the kth path. The received signal by the BTS can be written as (5)where is AWGN and is the transmitted signal by the radio transceiver. The power of the received signal varies randomly due to the multipath fading. The probability of bit error of the GSMK demodulator depends on the signal-to-noise ratio (SNR) of the received signal. In a fading environment, average SNR is more appropriate performance measure, which refers to statistical averaging over the probability distribution of the fading [19]. If denotes the instantaneous SNR at the GMSK receiver input that includes the impact of the fading of the radio channel, then (6) where expresses the probability density function (PDF) of , which is random depends on the variation rate of the fading. The fading of the signal level in the voice channel includes the effect of the fading of the radio channel. Fading phenomena can reduce voice signal strength. The GSM system employs power control to adjust the transmission power of the BTS and the mobile station (MS) to preserve the quality of the voice call and reduce the overall network interference in a time-varying mobile channel [20-22]. The BTS and MS measure the received signal quality (RxQual) and received signal level (RxLev) every 480 ms [23]. The transmission power will be increased or decreased according to the value of the RxQual and RxLev. However, it should be noted that the RxQual and RxLev are not accurate and real-time metrics for voice quality. The power control interval in the GSM voice channel is 480 ms. This means that the transmission power is fixed during each power control interval. Therefore, fading can affect a block of 720 bits in the fast modulation mode or 360 bits in the robust modulation mode. The existing power control technique in the GSM standard can keep the signal quality at an acceptable level for voice communication. However, the power control frequency is too slow for the MSD data transmission. Fading in the voice channel after power control increases the BER and packet loss of the MSD transmission. Particularly, the fading is strong in some regions with low signal strength and it can affect demodulation of the MSD packet from symbol to symbol even for a block of 720 bits during the 480 ms power control interval. Therefore, it is necessary to measure fading statistics of the GSM voice channel after power control and evaluate its impact on the MSD transmission. 3 Road measurements A testbed is designed and built to measure the fading characteristics of the voice channel in Fig. 2. The system consists of an IVS, the GSM network, the PSTN network, a PSAP transceiver and a computer to record the received signal. Data is processed after recording. Fig. 2Open in figure viewerPowerPoint Block diagram of the testbed for the EU eCall system, 3GPP TS 26.267/268/269 The IVS hardware platform contains a Freescale i.MX 6 board and a development board containing u-Blox GSM module LEON-G200 and GPS module LEA-6S with GPS antenna and GSM antenna in Fig. 3. The main processor of the Freescale i.MX 6 is capable of generating continuous wave (CW) or the eCall modulated MSD signal that are fed into a GSM voice codec. The output signal of the voice encoder is processed by the GSM baseband module and the radio module. The radio signal is transmitted through the GSM antenna and sent to a BTS of the GSM network. The signal received by the BTS is transmitted through the PSTN network and received by a telephone recording box located at the PSAP centre. A sound card inside the PSAP records the received signal. The sound card digitises the output waveform of the telephone recording box and the digitised waveform files are stored on a computer hard drive. The fading statistics of the received GSM voice signals are obtained by processing the recorded signals in Section 4. Fig. 3Open in figure viewerPowerPoint Experiment apparatus of the IVS Experiments have been carried out in different environments. The GSM radio signal is transmitted through T-mobile network. The carrier frequency of the GSM network uses 1900 MHz in the area. A vertical polarisation antenna was used in the test, and the antenna height is 15 cm. The frequency range of the antenna is from 700 to 2100 MHz. The antenna was sucked on the top of a Sedan with 145 cm height. The following test cases were studied. Case 1 Laboratory environment. The IVS was positioned in a laboratory surrounded by tall buildings. The experiment was conducted at the Advanced Communications Laboratory located on the third floor of the Engineering Building at Wayne State University. The IVS and GSM antenna were placed on a bench top in the laboratory. There was no line-of-sight propagation path between the IVS antenna and any cellular tower. The test in the laboratory lasted for 6 months for validation of the testbed. Case 2 Freeway with smooth traffic. The IVS was installed in a car running at high speed. The first experiment was performed between Exit 53 and Exit 83 of the Interstate-75 freeway from 9:30 a.m. to 11:30 a.m. on 26 March 2017. The IVS was installed inside a car. The GSM and GPS antennas were placed on the roof of the car similar to the GM OnStar antenna position. There was line-of-sight propagation path between the GSM antenna and the BTS. The test vehicle was running from south to north in the middle lane in cruise control mode at 70 mph. During the experiment, the traffic was smooth on the three-lane freeway. The freeway test was repeated in heavy traffic and the statistics were found similar to the freeway condition with smooth traffic. Freeway experiments lasted for more than two weeks to collect sufficient data of the eCall system. Case 3 Downtown Detroit. The first experiment was run in the Criswold Street and the Woodward Avenue in downtown Detroit from 2 p.m. to 5 p.m. on 11 April 2017. There are many 20-30 story buildings next to each other on both sides of the streets. There was no line-of-sight path between the IVS antenna and the BTS. The radio signal was affected by shadow fading, diffraction, scattering, multipath fading and other complications. The street was very busy with vehicles running and pedestrians. The speed of the testing vehicle was lower than 20 mph. The experiment continued for 2 days. Case 4 Rural road. The first experiment was performed on Wadhams Road, Kimball Township in Michigan from 1 p.m. to 5 p.m. on 17 April 2017. The driving speed was about 45 mph. Vehicles tend to move at higher speeds on rural roads than urban roads, but slower than on freeways. The GSM signal strength was very weak. The fatality rate on rural roads is highest among all traffic environments. More than 55% of fatal accidents occur on rural roads. The experiment continued for 2 days. The measurement procedure of the CW signal received by the PSAP is shown in Fig. 4. In order to measure the fading statistics of the voice channel, the amplifier gain was set to 0 dB for each of the IVS, the telephone recording box and the PSAP transceiver. The emergency button in the IVS was pressed to trigger an eCall. The PSAP answered the incoming call. After the voice call was established, the IVS sent a CW signal at a given frequency with 20 s frame which is the same length as the required maximum MSD transmission time. Meanwhile, the IVS saved the CW waveform into the SD card of the IVS so that it can be compared with the received waveform by the PSAP, and the PSAP recorded the received signal at 8 kHz sampling rate into 16-bit samples. This procedure repeats 500 times for each case. The saved data and waveforms are used to analyse the fading statistics of the GSM voice channel. Fig. 4Open in figure viewerPowerPoint Procedure to measure CW signal strength The measurement procedure of the eCall MSD is shown in Fig. 5. The IVS aggregated the MSD and initiated an eCall after the emergency button on the IVS was pressed. The PSAP sent a START command to the IVS. Then the IVS sent the MSD when the eCall was answered and a voice channel was established. The IVS and PSAP recorded the MSD waveform of each test sample. The data and waveforms were used to analyse the fading effect on the MSD success rate. Fig. 5Open in figure viewerPowerPoint Measurement procedure of the EU eCall MSD Experiments of the EU eCall system using the said testbed have been carried out for 5 years. Over 100,000 test calls were recorded. Data and waveform for the four test cases are analysed and discussed in the next section. 4 Fading statistics The signal received by the PSAP receiver is the sum of electromagnetic waves from different paths including reflection, diffraction and scattering [13, 14]. Power distribution of the received signal needs to be obtained through experiments for different environments, so that proper modulation and coding can be selected accordingly to increase the MSD success rate and minimise MSD transmission delay. Fading statistics for the EU eCall channel is also required to evaluate the performance of the MSD transmission including detection, synchronisation, modulation and coding. In this section, the CDF of the received signal power is obtained. LCR and AFD are found for the CW signals received by the PSAP in different test cases. The results are compared with a signal of Rayleigh distribution. Consider transmitting a CW signal through a mobile channel. The baseband signal include received in the in-phase channel and received in the quadrature channel, respectively. The signal envelope is (7)When and are Gaussian random variables with zero-mean and variance , is of Rayleigh distribution with PDF [24] (8)The corresponding CDF of the received signal envelope is (9) The instantaneous power of the received signal is . The average power of the received signal is (10)where is a time window for each calculation duration. Fig. 6 shows the PDF of the received CW signal power. The data was obtained from the test Case 2 with the transmitted CW signal at 500 Hz. Gaussian distribution and Rayleigh distribution with the same variance of the received signal power are plotted for comparison. It can be seen that the power distribution of the received CW signal is neither Rayleigh nor Gaussian. Fig. 6Open in figure viewerPowerPoint PDF of the received CW signal power versus Gaussian distribution and Rayleigh distribution. The data were obtained from the test Case 2 with the transmitted CW signal at 500 Hz The LCR is defined as a rate at which the received signal envelope crosses a given level Z in the downward direction after the fading envelope is normalised to the root mean square (RMS) level [24, 25]. The number of crossings of the received signal envelope level Z over the interval [0, T] is (11)where is the derivative of versus time, is the joint PDF of z and at , is the maximum Doppler frequency and is the value of the specified level Z which is normalised to the RMS level of the fading envelope. The LCR is (12)which is the number of crossings of the envelope level Z per second. The AFD is the average time that the signal envelope stays below a given target level Z in the normalised signal amplitude [26]. Let denote the duration of the ith fade below level Z over a time interval . Let (13)The AFD is (14) The AFD also can be expressed as a function of and , which is (15) The AFD of a faded signal helps to determine the number of data bits that may be affected during fading. Fig. 7 shows a 10 ms segment of CW signal received by the PSAP. It contained three deep fading durations , and below dB fading depth. The fading depth is the reduction from the RxLev to the RMS of the signal measured in dB. The fades fell below dB on three occasions in 10 ms. One of these fades was even lower than dB. According to the LCR definition, the LCR of the signal is three crosses per 10 ms (or 300 crosses/s). The AFD is ( + + )/3, which is about 0.4 ms when the signal power stays below the threshold dB. When a deep fade exists in the channel, the PSAP is likely to fail in demodulating and decoding the MSD. In this case, the PSAP centre sends the negative acknowledgement feedback message to the IVS and requests to resend the data packet. Therefore, the signal fading can increase the MSD transmission delay and even lead to MSD data transmission failure. Fig. 7Open in figure viewerPowerPoint Segment of the RxLev for a GSM network. The RMS of the signal level is 0 dB. The signal crossed dB for three intervals Fig. 8 shows the MSD signal sent by the IVS and the received MSD signal by the PSAP. Fig. 9 shows deep fade existed at the start of the received CW signal. There was deep fade in the time interval [40, 500] ms. The synchronisation tone of the EU eCall is a 64 ms sinusoidal tone of frequency 500 Hz for the fast modulation mode. It is suggested that the synchronisation tone in the 3GPP TS 26.267 to be replaced by a synchronisation tone of 600 ms at 2000 Hz for reliable synchronisation. Fig. 8Open in figure viewerPowerPoint Original MSD signal sent by the IVS and the received MSD signal by the PSAP Fig. 9Open in figure viewerPowerPoint Segment of the signal received by the PSAP. The signal transmitted by the IVS was CW at 500 Hz. Deep fade existed at the start of the received signal Fig. 10 shows the CDFs of the CW signals at 500 Hz received by the PSAP in different test environments. The Rayleigh distribution is plotted for comparison. The data was collected in the laboratory, on the freeway, urban road and rural road. The fading cumulative distributions were compared to the Rayleigh distribution. When the probability was less or equal to , the fading and attenuation was dB on the urban road, dB on the rural road, dB on the freeway and dB in the laboratory, respectively. The maximum fade occurred on the urban road. When the probability was less or equal to , the fading and attenuation was dB on the urban road, dB on the rural road, dB on the freeway and dB in the laboratory, respectively. When the received signals were 10 dB below the median value, the probability was on the rural road, on the urban road, on the freeway and in the laboratory, respectively. The results show that the signals received by the PSAP exist strong fading, which will increase the data packet loss rate and reduce the data transmission success rate. Fig. 10Open in figure viewerPowerPoint CDFs of the CW signals at 500 Hz received by the PSAP in different test cases. The Rayleigh distribution is plotted for comparison Fig. 11 shows the CDFs of the CW signals received by the PSAP at different frequencies on the I-75 freeway. The Rayleigh distribution is plotted for comparison. When the probability was less or equal to , the fading and attenuation was dB for the 500 Hz CW signal, dB for the 800 Hz CW signal, dB for the 1000 Hz CW signal, dB for the 1500 Hz CW signal, dB for the 2000 Hz CW signal, dB for the 2500 Hz CW signal, dB for the 3000 Hz CW signal, dB for the 3500 Hz CW signal, respectively. When the received signals were 10 dB below the median value, the probability was for the 500 Hz signal, for the 800 Hz signal, for the 1000 Hz signal, for the 1500 Hz signal, for the 2000 Hz signal, for the 2500 Hz signal, for the 3000 Hz signal, for the 3500 Hz signal, respectively. It can be seen that the fading of the 500 Hz CW signal and 800 Hz CW signal were more severe than the fading of the 1500 Hz CW signal and 2000 Hz CW signal. The worst fading oc
Referência(s)