Experimental Wireless Ultra Wideband Sensor Network ... - IEEE Xplore

1 downloads 0 Views 263KB Size Report
Abstract— In this paper the experimental UWB wireless sensor data collection network is ... The UWB signals are often used in data collection systems [1–6].
2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

Experimental Wireless Ultra Wideband Sensor Network for Data Collection Sergey V. Volvenko1 , Dong Ge2 , Sergey V. Zavjalov1 , Alexander S. Gruzdev1 , Andrey V. Rashich1 , and Evgeniy L. Svechnikov1 1

Peter the Great St. Petersburg Polytechnic University, Russia School of Aerospace Engineering, Tsinghua University, China

2

Abstract— In this paper the experimental UWB wireless sensor data collection network is presented. The simulation of reception algorithm was performed using Matlab. The number of UWB transmitters was 4. Packets of different transmitters had different formats. It is shown that the increasing number of repetitions on the physical layer allows achieving BER performance close to theoretical limit. The increasing number of repetitions also provides increased probability of packet delivery. The length of preamble is 1000 elements and is defined by the requirements of minimization of BER and maximization of probability of packet delivery. Functional tests conducted for the worst case of simultaneous operation of all transmitters. 1. INTRODUCTION

The UWB signals are often used in data collection systems [1–6]. UWB data collection networks can use either synchronous or asynchronous data transmission technologies. It is well known [7–9] that synchronous UWB receivers are technically much more complex than asynchronous ones. In asynchronous receivers, a combination of energy detection and “On-Off Keying” (OOK) modulation is often used [1, 7–11]. Various reception methods are proposed for the signals with OOK modulation [1, 7, 8]. Let us note that the OOK modulation is the most energy efficient transmission method. The main drawback of this modulation method is its inability to provide high speed communication. During previous work which was performed under contract with AO “Klimov” the requirements to parts of the wireless data collection system were defined including the types and numbers of sensors, requirements to a UWB transmitter, receiver and signal processing subsystem. These requirements were used during development of an experimental wireless sensor network. The work objective was to develop transmitter and receiver modules for experimental UWB wireless sensor network. Data sources are different sensors installed on the engine. The total number of transmitter modules is 4 including 3 transmitter modules for thermal sensors and 1 transmitter module for thermal, rotation frequency and pressure sensors. Each transmitter module uses a unique m-sequence. The length of the sequence is 32 elements. Frames with 32Nb elements (Nb — number of transmitted bits) are formed by applying the msequence to data bits. The formed frames are transferred to the network physical layer. A preamble is then added to each data frame. M -sequence is used as a preamble on the physical layer. Preambles are different for different transmitter modules. Physical layer packet format is presented in Fig. 1. A completed physical layer packet is transferred to the OOK modulator. Data link frame (max 61 bits) Data max 49 bits Nb bits

ID 4 bits

CRC 8 bits

→ Spread sequence application (32 elements)



Physical layer packet Preamble L elements

Data 32Nb elements (L + 32Nb) elements

Figure 1: Physical layer packet and data link frame. 965

2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

Each data frame has a packet type identifier (ID) and a check sum (CRC-8). Possible data types are: 1) “Temperature data #1–3” (9 bits). This data must be transferred every 2.5 ms. The following data types are formed by a data collection and processing unit. 2) “Pressure data #1” (8 bits). This data must be transferred every 2.5 ms. 3) Common data: “Pressure #1” (8 bits), “Frequency #1” (11 bits), “Frequency #2” (12 bit). This data must be transferred every 25 ms. 4) Common data: “Pressure #1” (8 bits), “Frequency #1” (11 bits), “Frequency #2” (12 bits), “Temperature #4” (9 bits), “Pressure #2” (9 bits). This data must be transferred every 100 ms. Note, that the duty cycle for UWB signals is usually small. This allows simultaneous operations of several transmitters with Aloha as channel access method. In this case the simultaneous transmission by several transmitters is the worst case scenario. It leads to the loss of information from all transmitters. However, the probability of such collision is relatively small. Repetition of each physical layer packet of up to 4 times with a random pause between repetitions is used to prevent collisions. 2. SIMULATION MODEL

Simulation model was developed in Matlab for preliminary evaluation of the proposed system feasibility. Block diagram of the model is shown in Fig. 2(a). The generation of transferred data and checksum calculation are performed in the first step. Spreading sequence application and addition of preamble are performed in the next step. Generated physical layer packet (Fig. 1) is passed to the OOK modulator. The resulting UWB signal is transmitted through an AWGN channel. Mixture of UWB signals with noise is sent to the block “Energy Detection” at the receiver. This block simulates the analog part of the receiver and is a thresholding device. Output values are “+1” if the signal exceeds the threshold value and “−1” otherwise. The output values from the “Energy Detection” block are transferred to the preamble search unit (Fig. 2(b)). The packet identifier processing and data reception are performed if preamble is found. As the last step the checksum of the received data is calculated. The data is considered correct if the calculated check sum is equal to the received check sum. Note, that in the simulation model there are 4 transmitters for “Temperature data #1–3” and 1 data collection and processing unit. Preamble search

Data Array of received samples

Received data Calculating and adding a checksum

Verifying a checksum

Signal from Energy Detection unit {±1}

Shift array Record the received value to the array

Spread sequence application

x

Preamble

Packet Receiver The adder over preamble length

Adding a preamble Preamble search

Preamble found / Preamble not found

Decision unit

Energy Detection

UWB modulator

(a)

Threshold

(b)

Figure 2: (a) Block diagram of the model; (b) preamble search unit.

Figure 3 shows the simulation results. The preamble in this case is 1000 elements long. Analysis of the simulation results confirms that the proposed UWB sensor network can be created. It must be taken into account that the energy gain of 12 dB over BER performance of simple OOK signals is achieved by using the spreading sequences. It can be seen that 3 repetitions can produce BER performance that is close to the potentially achievable and allows reaching the delivery probability that is close to 1. Therefore, there is no need to repeat the data more than 3 times. 966

2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

–2

0.95 Delivery probability

10 Error probability

1

without repetition 1 repetition 2 repetition 3 repetition

10–1

–3

10

10–4

0.9 0.85 0.8 without repetition 1 repetition 2 repetition 3 repetition

0.75 0.7

–5

10

0.65

–6

10

–3

–2 –1 0 Signal-to-noise ratio, dB

1

0.6 –3

–2 –1 0 Signal-to-noise ratio, dB

(a)

1

(b)

Figure 3: Simulation results. 3. EXPERIMENTAL SETUP

An STM32F401 microcontroller was used in the UWB transmitter. A K-type thermocouple can measure temperatures from −60 to +960◦ . The measured temperature values are transferred to the microcontroller via the SPI serial interface. The UWB pulse generator uses a 2D524A step recovery diode. The generator’s schematic can be found in [12]. The generator’s trigger pulses are generated using another SPI interface. The period of the trigger pulses is 1 μs.

SPI Thermal sensor

SPI STM32F401

UWB impulse generator

Figure 4: Transmitter structure.

The UWB pulses from the generator are sent to the UWB antenna. The transmitting and receiving antennas are UWB monopoles with shields. Each antenna occupies 60 × 50 × 20 mm3 volume. The antennas provide radiation into one hemisphere and their performance is practically independent on the shape of an object on which they are installed (Fig. 5(a)). These properties allow easy antenna installation on various objects.

(a)

(b)

Figure 5: Antenna drawing (a) Antenna VSWR vs. frequency (b).

The voltage standing wave ratio (VSWR) of the antenna was measured using an HP8510 vector network analyzer. The VSWR is less than 2.2 over the 1.5 to 10 GHz frequency band. The radiation pattern width is less than 30 degrees in both planes up to at least 5 GHz. The wide bandwidth 967

2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

allows radiation of pulses of various shapes and durations. On the other hand, such antennas can potentially be adapted to the spectral properties of a specific waveform in order to achieve the maximum peak amplitude of the transmitted or received pulses. In order to test the transmitter modules and investigate the feasibility of creation of receiver modules, the signals from 4 different transmitters were recorded. Each transmitter had a unique preamble (1000 elements long) and spreading sequences. The distance between the transmitters and the receiver was 2 meters. The duration of each recording was 10 ms. The example of a recorded signal (the packet containing the “Temperature #1” value) is shown in Fig. 6. The recording contains 4 repetitions of a packet with random intervals between them. The signals received by the antenna were recorded using an Agilent Technologies DSO9104A oscilloscope at the rate of 1 Gs/s.

Figure 6: An example of the transmitter packages record.

The recordings were summed element by element. This procedure simulated the worst case of the minimum time shifts among the transmitted signals. The resulting mixture of signals was sent to the input of the Matlab simulation model (Fig. 2(a)) that performed the signal processing. The experimental results are presented in Table 1. The reception was considered correct if the calculated and received checksums matched. As it can be seen at least one repetition of the data packets sent by each transmitter was received and processed correctly. Table 1: Experimental results. Transmitter number The number of correctly received packets

1 4/4

2 2/4

3 1/4

4 2/4

Thus the possibility of creating a receiving module operating according to the algorithm shown in Fig. 2 was demonstrated and the operation of the transmitter modules was checked. The single channel receiving module contains a mixed analog-digital part and an FPGA based digital part (Fig. 7). The signal received by an antenna is amplified by about 15 dB by a lownoise amplifier (LNA). The amplified signal is sent to the energy detector (ED) block where it is compared with a reference level using a high-speed analog comparator. After the comparator a

LNA

ED

D-trigger

FPGA

Figure 7: Block diagram of a receiver. 968

UART/USB

2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

D-trigger generates a rectangular pulse. The duration of this pulse is defined by the reset signal from the FPGA. The D-trigger generates two output signals with TTL levels that are converted to the ECL levels by a level convertor. The level converter output is connected to the FPGA input. Processed data are sent to the PC via an USB based UART. start bit_in reset_comparator

Timing synchronization block

channel bits

Detector

channel bits

Spread sequence demodulator

info bits

Check-sum calculator

info bits clk100Mhz

125 MHz clock generator Reset generator

clk125Mhz

UART transmitter

srst

data to host PC

Figure 8: FPGA part of UWB receiver.

Figure 9: Received data packet as seen on the oscilloscope’s screen.

The digital signal processing module structure (Fig. 8) follows the algorithm shown in Fig. 2. The “Timing synchronization” block is an important part of the receiver module. This block receives signals from the D-trigger and resets it using reset comparator signal. The repetition period of data bits in the channel is 1 μs. FPGA must also generate the signal reset comparator with 1 μs period. But because the frequencies of the clock generators in the transmitter and receiver are not equal the period of the transmitted bit sequence is also not equal to the reset comparator signal period. This results in lower quality of reception due to timing errors. The “Timing synchronization” block analyses the time difference between the channel bits and the reset comparator signal. If the difference exceeds a certain limit the repetition period of the reset comparator signal is corrected. After passing the “Timing synchronization” block the channel bits are transferred to the “Detector” block. The preamble detection block called the “Detector” use correlation detection of the packet’s preamble and generates the trigger pulse for the data bits (coded using spreading sequences) processing blocks called the “Spread sequence demodulator” and the “check sum calculator” block. The “Detector” block consists of a matched filter and a solver. The matched filter is implemented using the serial-parallel folded architecture. The solver compares the absolute value of the correlation coefficient at the output of the matched filter with a threshold value. If the value exceeds the threshold then the trigger signal for the next signal processing blocks is generated. The “Spread sequence demodulator” block converts the channel bits to information bits. The conversion is done by counting the number of coincidences of the channel bit values with the corresponding bit values in the 32-bit long spreading sequences. The solution about the value of the information bit is made using the majority principle. 969

2017 Progress In Electromagnetics Research Symposium — Spring (PIERS), St Petersburg, Russia, 22–25 May

The data integrity is checked using the 8-bit CRC-8 Dallas/Maxim checksum. The generator polynomial is x8 + x5 + x4 + 1. The “check sum calculator” block is started by the command from the “Detector” block and calculates the checksum of the received information bits. The checksum is computed bitwise, the checksum register is initialized by the value “11111111”. The information bits, checksum, checksum flag and the correlation coefficient value are sequentially sent to the “UART transmitter” block. The UART block uses FTDI serial to USB convertor chip to send the data to a PC. As a demonstration, Fig. 9 shows the received data for the packet containing the “Temperature #1” parameter. During the transmission, the ID field contained the value “0001”, the data field contained the value “001000110” and the checksum value was “10000110”. As can be seen the transmitted and received bit sequences are equal. 4. CONCLUSIONS

The experimental wireless UWB sensor network was developed. Future research will be directed towards investigation of the possibility of using shorter preambles and spreading sequences. Details of the transmitting and receiving modules design will be presented in more detail in the future publications. ACKNOWLEDGMENT

The results of the work were obtained under the State contract No. 8.1873.2014/K with Ministry of Education and Science of the Russian Federation and used computational resources of Peter the Great Sainte-Petersburg Polytechnic University Supercomputing Center (http://www.scc.spbstu.ru). REFERENCES

1. Ghavami, M., L. B. Michael, and R. Kohno, Ultra-wideband Signals and Systems in Communication Engineering, xxviii, 247, John Wiley & Sons, Chichester, 2004. 2. Jin, Y., et al., “Cyclic prefixed single carrier transmission in intra-vehicle wireless sensor networked control systems,” 2014 IEEE 79th Vehicular Technology Conference (VTC Spring), 2014. 3. Dai, X., et al., “Wireless communication networks for gas turbine engine testing,” International Journal of Distributed Sensor Networks, 18, 2012. 4. Bas, C. U. and S. C. Ergen, “Ultra-wideband channel model for intra-vehicular wireless sensor networks beneath the chassis: From statistical model to simulationsIEEE Transactions on Vehicular Technology, Vol. 62, No. 1, 14–25, 2013. 5. Slottke, E., et al., “UWB marine engine telemetry sensor networks: Enabling reliable lowcomplexity communication,” 2015 Ieee 82nd Vehicular Technology Conference (Vtc Fall), 2015. 6. Ameti, A., et al., “Ultra wideband technology for aircraft wireless intercommunications systems (AWICS) design,” IEEE Aerospace and Electronic Systems Magazine, Vol. 19, No. 7, 14–18, 2004. 7. Witrisal, K., et al., “Noncoherent ultra-wideband systems,” IEEE Signal Processing Magazine, Vol. 26, No. 4, 48–66, 2009. 8. Wang, F., Z. Tian, and B. M. Sadler, “Weighted energy detection for noncoherent ultrawideband receiver design,” IEEE Transactions on Wireless Communications, Vol. 10, No. 2, 710–720, 2011. 9. Mu, D. Z. and Z. D. Qiu, “Weighted non-coherent energy detection receiver for UWB OOK systems,” ICSP: 2008 9th International Conference on Signal Processing, Vol. 1–5, 1847–1850, 2008. 10. Cheng, X. T., Y. L. Guan, and S. Q. Li, “Optimal BER-balanced combining for weighted energy detection of UWB OOK signals,” IEEE Communications Letters, Vol. 17, No. 2, 353– 356, 2013. 11. Cheng, X. T., Y. L. Guan, and Y. Gong, “Thresholdless energy detection for ultra-wideband block-coded OOK signals,” Electronics Letters, Vol. 44, No. 12, 755–756, 2008. 12. Protiva, P., J. Mrkvica, and J. Machac, “Universal generator of ultra-wideband pulses,” Radioengineering, Vol. 17, No. 4, 74–78, 2008.

970