Fully digital intensity modulated LIDAR

Faio POLLASTRONE*，Gian Carlo CARDARILLI，Roerto PIZZOFERRATO，Maro RE

aDepartment FSN，F(xiàn)USTEC-IEE-ENEA Frascati，Roma，Italy

bDepartment of Electronics Engineering，University of Rome “Tor Vergata”，Roma，Italy

cDepartment of Industrial Engineering，University of Rome “Tor Vergata”，Roma，Italy

Fully digital intensity modulated LIDAR

Fabio POLLASTRONEa，*，Gian Carlo CARDARILLIb，Roberto PIZZOFERRATOc，Marco REb

aDepartment FSN，F(xiàn)USTEC-IEE-ENEA Frascati，Roma，Italy

bDepartment of Electronics Engineering，University of Rome “Tor Vergata”，Roma，Italy

cDepartment of Industrial Engineering，University of Rome “Tor Vergata”，Roma，Italy

In several applications，such as collision avoidance，it is necessary to have a system able to rapidly detect the simultaneous presence of different obstacles.In general，these applications do not require high resolution performance，but it is necessary to assure high system reliability also within critical scenarios，as in the case of partially transparent atmosphere or environment in presence of multiple objects （implying multiple echoes having different delay times.）This paper describes the algorithm，the architecture and the implementation of a digital Light Detection and Ranging（LIDAR）system based on a chirped optical carrier.This technique provides some advantages compared to the pulsed approach，primarily the reduction of the peak power of the laser.In the proposed architecture all the algorithms for signal processing are implemented using digital hardware.In this way，some specif i c advantages are obtained:improved detection performance （larger dynamics，range and resolution），capability of detecting multiple obstacles having different echoes amplitude，reduction of the noise effects，reduction of the costs，size and weight of the resulting equipment.The improvement provided by this fully digital solution is potentially useful in different applications such as:collision avoidance systems，3D mapping of environments and，in general，remote sensing systems which need wide distance and dynamics.

CW-IM LIDAR；Chirp；Laser Obstacle Avoidance

1.Introduction

LIDAR based on laser beam scanning can be applied to several detection and ranging f i elds，including obstacle avoidance in aerospace navigation ［1］as well as real-time surveillance of restricted areas.For example，LIDAR can be used in port areas security to detect crafts in rapid approach，which are not easily revealed by passive optical systems at night，also considering that RF Radar systems can fail in case of nonconductive or small boats.

Many laser modulation techniques can be applied，obtaining different measurement ranges and resolutions:

1）Continuous wave amplitude modulated ［2］，based on the sinusoidal modulation of the laser beam intensity （submillimetrical resolution，single echo and small distance）；

2）Pulsed LIDAR ［3］（long distance，multiple echoes）；

3）Pulse compression ［4，5］（long distance and multiple echoes）

4）Continuous wave frequency modulated （CW-FM）technique （long distance and multiple echoes）［6，7］.

5）Continuous wave intensity modulated ［8］，based on the laser beam intensity modulated by a chirped signal.

The CW-IM-technique is generally implemented by using analog electronic circuits or optical system.Despite its greater operative frequency which can allow a higher resolution，the use of an analog implementation reduces the f l exibility and the robustness of the obtained equipment，and does not enable the application of powerful processing techniques that can improve the performance in presence of multiple echoes with very different amplitudes.On the contrary，a digital approach is able to exploit these techniques，increases the integration and reduces the complexity of the assembling ［9］.As a consequence，the resulting devices have reduced costs and increased reliability.

For these reasons during the last years，the authors developed different versions of a fully digital processing system for LIDAR.This system is able to measure the times of f l ight of the optical wave also in presence of multiple echoes.The system has been developed in a collaboration between ENEA and University of Rome “Tor Vergata”.

All the above versions of the electronic circuits for the LIDAR have been designed and tested；one of these has also been actually applied to an optical laser probe.

This paper describes the CW-IM algorithm used in the experimental equipment，the architecture of the hardware and f i rmware developed，the test performed and their results.

The paper is organized as follows:in Section 2 the algorithm is brief l y discussed，while in Section 3 the architecture of the fast prototype is illustrated.Section 4 contains a discussion on the digital implementation of the proposed algorithm.Section 5 describes the experimental results of a f i rst version of the LIDAR electronic system，while Section 6 contains the preliminary electrical test of the second release.The last section contains the conclusions and the possible future activities.

2.CW-IM algorithm

CW-IM LIDAR technique is based on linear complex chirp signal

where fr= Δfreq/T is the increasing rate of chirp frequency，T is the sweep duration and Δfreq=stop_freq-start_freq is the chirp bandwidth.

The laser beam is modulated with the component

The echo signal S（t）at the output of the photodiode that receives the lights backscattered from the targets is given by

S（t）corresponds to RQ（t）delayed by time of fl ight Δt=2·D/C（where D is the target distance and C is the speed of light）.The amplitude of the echo ARdepends on the target material，the angle of incidence and the distance.

For sake of simplicity，in this discussion additional phase shifts in the echo have not been considered；this assumption does not affect the fi nal results.

The product C（ t） = R（ t）·S（ t ）can be expressed as

where

The high-frequency terms CIHand CQHare 2 chirps with double chirp rate and different start and stop frequency with respect to R（t）.The most of CIHand CQHsignals are removed from the C（t）signal using a complex low-pass f i lter，as shown in Fig.1.The remaining low-frequency terms CILand CQLare 2 sinusoids at frequency2 frΔt ，they depend on the time of fl ight Δt.

Fig.1.Trend of the signals frequencies during the chirp period.

The instantaneous frequencies of the previously described signals are shown in Fig.1.

In case of multiple echoes with different delays Δti，it is possible to know the amplitude ARiof the single echo by analyzing the module of the Fast Fourier Transform （|FFT|）of the（CIL+i CQL）signal.

3.Architecture of the fast prototyping system

Two fully digital CW-IM LIDAR electronics have been developed starting from Field Programmable Gate Array（FPGA）fast prototyping system ［10］.

The f i rst release （see Fig.2）is based on Stratix II EP2S60 DSP Development Board presenting the following characteristics:

1）Altera Stratix II EP2S60F1020C4 FPGA；

2）100 MHz system clock；

3）Two 12-bit 125 MspsA/D （modelAD9433BSQ）converters used in interleaved mode to obtain a 150 Msps analog to digital conversion；

4）14 bit 165 Msps D/A converter （model TI DAC904）；

5）An Ethernet MAC/PHY；

6）A JTAG interface.

Moreover，the hardware contains a signal conditioning circuitry and an optical interface composed by a laser diode and a photoreceiver.

4.Implementation of CW-IM LIDAR algorithm on digital hardware

The proposed algorithm has been implemented through the digital processing of the signals；the main limitations are due to the sampling frequency （Fs）of the A/D converter.If compared with the conventional analog implementations we obtain the following advantages:

1）Simplif i cation of the system and lower cost due to the absence of critical analog parts；

2）The digital generation of the quadrature complex chirp signal（with frequency in the range 0-Fs/2）corresponds toa better stability/linearity and f l exibility in comparison with the analog implementations；

Fig.2.First version of the fully digital CW-IM LIDAR electronic system.Fast prototyping board architecture.

3）High linearity of the complex product and of the digital low pass band f i lters （FIR）；

4）Small size and very low weight；

5）Flexibility due to the possibility to reprogram the FPGA（also in real time）by loading new algorithms；

6）Scalability，with the proposed approach we can take full advantage ofthe technologicaldevelopments； for example A/D speed improvements can be exploited to increase the resolution；

7）The digital implementation inside the FPGA of the complex chirp R（t）makes it possible to use a reference signal that can be used directly without any A/D conversions.This characteristic permits to use greater wordlengths for the chirp representation，allowing a signif i cant optimization of the algorithm.

The f i rst implementation uses two A/D converters in interleaving mode for increasing the input sampling frequency （up to 150 Msps）.The performance degradation due to sampling jitter and linearity mismatches of the two A/D converters are acceptable in our application.However，methods for correcting these errors are already present in the literature.

The LIDAR electronic hardware/f i rmware architecture is shown in Fig.3.The f i rmware has been implemented inVHDL by using ALTERA Quartus II Macrofunctions.

The FPGA f i rmware is composed of the following blocks:

1）The inter leaving mux that alternatively selects the S1 and S2 signal，outputting the S（t）echo signal@150 Msps；

2）One Quadrature Chirp Generator that generates the R（t）reference modulation chirp （20 bit resolution）.The Quadrature Chirp Generator is composed of a Numerically Controlled Oscillator （NCO）combined with a Ramp_generator.TheRamp_generatoroutputsthe Phase_Increment （proportional to the instantaneous frequency）and the Reset of the NCO and is controlled by the Chirp_Sync.The Chirp_Sync generates the Start_Ch and Stop_Ch signals that command the increase and the reset for the Ramp_Generator；

3）A Complex Multiplier implementing the product of the echo signal S（t）and the complex modulation signal R（t）；

Fig.3.Hardware/f i rmware for implementation of the algorithm.

4）A low pass decimator FIR f i lter implemented by using the Altera FIR compiler tool；the low pass decimator FIR is composed of two decimator FIR in cascade and the FIR characteristics are reported as the following:

F i r s t F I R S a m p l i n g f r e q u e n c y （Fs1） 1 5 0 M H z D e c i m a t i o n -f a c t o r 4 T a p s 5 0 W i n d o w t y p e H a m m i n g -3 d B f r e q u e n c y Fs1× 0 . 0 4 -5 5 d B f r e q u e n c y Fs1× 0 . 1 S e c o n d F I R S a m p l i n g f r e q u e n c y （Fs2） Fs1/ 4 = 3 7 . 5 M H z D e c i m a t i o n f a c t o r 2 T a p s 6 0 W i n d o w t y p e B l a c k m a n -3 d B f r e q u e n c y Fs2× 0 . 1 -6 0 d B f r e q u e n c y Fs2× 0 . 1 8 = 6 . 7 5 M H z

5）A f l oating point FFT （2048 samples），synchronized with the R（t）chirp modulation start，implemented by using the Altera FFT IP Core.

A PC，connected as shown in Fig.2，is used to conf i gure the FPGA board and to acquire the elaborated signals （such as the FFT result）.The connection PC-FPGA board is realized by using the USB blaster device and the Altera software tools.

In the following，some considerations about the frequency limitations related to the digital implementation of the detection algorithm are reported.As shown in Fig.1 （see curve R（t）），it is possible to generate and sample signals with instantaneous frequency up to Fs/2=75 MHz.Moreover，aliasing phenomena are possible in the high frequency components CIHand CQHof the product signal C（t）.

The LIDAR resolution can be improved by increasing the sweep bandwidth and the chirp rate fr.For this reason and considering the hardware constraint，in the f i rst version of the LIDAR electronic realization the frequency of R（t）has been swept from 0 MHz to 75 MHz.The low pass FIR decimator（decimation factor=8；cut-off frequency 0.35·Fs/8 ＜ ?·Fs/ decimation_factor）is needed to remove most of the CIHand CQHcomponents.However，being CIHand CQHtwo chirp signals，during the chirp period there are two short time intervals where residual signal are present，not removed by the f i lters （see yellow triangles in Fig.1）.The reduction of the FIR cut-off frequency decreases the duration of CIHand CQHresiduals，improving the signal to noise ratio （SNR），but on the other hand reduces the maximum measurable range.

Taking into account the decimation factor and to avoid discontinuity of the CLcomponents （due to an FFT window longer than the R（t），with the resulting presence of two or more sequencesR（t）inthewindow）thenumberofsamplesfortheFFT windowislimitedto2048.Thechirplasermodulationperiodhas been set to 16，384 samples （corresponding to 109 μs）.The characteristics of the used electronic board limit the resolution，because it depends on many parameters （start_freq，stop_freq，sampling frequency Fsetc.）.In our f i rst implementation setting the start_freq=0 and stop_freq=Fs/2 we obtain

Table 1Main parameters related to f i rst fully digital LIDAR electronic developed.

On the other hand，it is possible to extend the measurement range by increasing the sweep_duration or reducing the decimation_factor，or by using a more sophisticated laser modulation based on phase-shift keying ［11］.The maximum detectable frequency is limited by the cut-off frequency of the FIR decimation f i lter；in our case，F(xiàn)s× 0.35/（decimation_factor ×0.5）is the available output bandwidth.As a consequence，the theoretical range obtained in the developed system is

Table 1 summarizes the main parameters related to the f i rst version of the Fully Digital LIDAR system.

The measurement throughput （? 10kHz）allows low frame rate imaging.

Table 2 reports the Main f i rmware functions of the CW-IM algorithm.In particular the resource utilization is related to the f i rst version of the LIDAR electronics （FPGA Altera 2S60）.

5.System test

The system has been tested by using two different approaches:a）by emulating the optical delay with an electrical delay line；b）by using a laboratory optical set up.Moreover，the experimental results （reported below）were compared and found in agreement with preliminary simulation using Matlab and Simulink，

5.1.Electrical test

An electrical test has been performed on the f i rst version of the LIDAR electronics in order to verify its behavior in the caseof various types of signals in input to the system.In particular，the electrical characterization of the digital LIDAR system considers different attenuation and delay times for the echo signal.

Table 2Main f i rmware functions and relative utilization of the FPGA resources （f i rst LIDAR electronics Altera 2S60）.

The experimental set up is composed of a coaxial cable in series with a variable attenuator that sends back the output of the modulation signal RQ（t）to the input S（t）signal （Fig.4）. Moreover，a FIFO has been implemented in the FPGA for emulating greater echo delays.

The test carried out on this f i rst version of the LIDAR electronic prototype demonstrates the correctness of the time delay measured，and the linearity of the echo amplitude.

Fig.4.Electrical test layout for the f i rst LIDAR electronic prototype.

Fig.5.Electrical tests of the f i rst LIDAR electronic prototype.|FFT|for different echo attenuation.

Fig.6.Distance measure obtained with echo signal attenuated by 72 dB.

In particular，as shown in Figs.5 and 6 it is possible to detect echoes with attenuation up to 72 dB，although in this f i rst embodiment it is not possible to discriminate echoes of very different amplitudes.This is due to the fact that the greater echo，through its residues，introduces a background noise that covers the smaller echo.This problem has been reduced in the second LIDAR electronic （see Section 6）.

The target distance is calculated applying the equation:

where#sample|FFT|peak represents the bin number （FFT output sample）having the maximum FFT magnitude.The constant dist_offset is due to the hardware component delays（LIDAR electronics，laser，photodetector，cables etc.）.

5.2.Optical test

A more sophisticated test bed has been developed by including a simplif i ed Optical Laser Probe，in order to test the above implementation.

In this case the electrical signal RQ（t）has been connected to the modulation input of a 75 mW CW 660 nm semiconductor laser diode （Melles Griot Mod.56RCS008/HS），and S（t）is obtained from the electrical output of a photodiode （Thorlabs Mod.PDA10A-EC）with an optical bandwidth from 200 nm to1000 nm and a RF output bandwidth from DC to 150 MHz.The experimental set-up is shown in the picture of Fig.7.

Fig.7.Photograph taken during optical tests.

Fig.8.Optically-measured|FFT|echo distance versus actual target distance.

Fig.9.Optically-measured FFT Module in a double echo case.

Fig.8 shows the relation between the measured positions of the obstacle （using FFT magnitude）and the actual target distance；the linearity error is very low in comparison with the theoretical resolution （2 meters）.

Fig.9 shows the output of FFT in the case of presence of two different echoes，the f i rst（15 meters）is related to a transparent glass slide and the second is related to the target （54 meters）. Both pulses are easily detectable.

6.Upgrade of the LIDAR electronics end relative tests

The CW-IM digital LIDAR electronics has been recently upgraded to a new version with a more performant proto board. It is based on:

Altera Stratix 3SL150 FPGA （142K logic elements）

two 14 bit 150 MSPS A/D converters

two 14 bit 250 MSPS D/A converters

Considering the increase of the sampling frequency of the A/D converters （overclocked to 165 Msps），the new architecture does not need the interleaving features.Moreover，while the 2 meter distance resolution has been conserved，minor improvements in the algorithm implementation （start_freq，stop_freq，sweep_duration）have been made.In particular，by increasing the out-of-band rejection of the low-pass FIR and the start_freq，and reducing the stop_freq，the CHresiduals have been drastically reduced.

Fig.10.Electrical tests of the second LIDAR electronic prototype:spectrum vs echo attenuation.

Fig.11.Electrical test:comparison of the SNR obtained using f i rst and new hardware architecture.

The result is a decrease of the signal noise f l oor （see Fig.10）and a consequent improvement of the Signal to Noise Ratio（SNR），particularly in the case of high level echo signals （see Fig.11）.The reduction of the noise level increases the ability to detect small amplitudes echoes，while the increase of SNR allows the detection of simultaneous electrical echoes having very different amplitudes.

Fig.11 shows the comparison of the SNR between the f i rst and the new version of the LIDAR electronics.On the graph the SNR is represented in relation to the normalized echo input amplitude.

In the f i rst version of the LIDAR electronic system the SNR saturates at about 43 dB，while in the new system the SNR increases fairly linearly with the echo amplitude from 35 dB （in the case of input attenuation of 72 dB）up to 117 dB in the case of full scale echo （0 dB attenuation）.

7.Conclusions

In this paper the digital implementation of a LIDAR based on a CW-IM laser modulation has been presented.If compared with the analog implementation，this approach gives interesting advantages in terms of cost，performance，f l exibility and physical size of the f i nal equipment.The proposed architecture is based on the FFT which allows a very eff i cient implementation of the algorithm.The relation between the different algorithm parameters and the system performance has been analyzed. Moreover，the algorithm has been tested implementing two hardware prototypes.These prototypes have been used for different experiments enabling the evaluation of the performance of the whole equipment.All the results show that this technique is suitable to implement an eff i cient low-cost LIDAR，particularly useful for defense and security applications.The improvements obtained in the second version of LIDAR electronic prototype are particularly interesting in terms of decrease of RNR.Optical tests of the second LIDAR electronic version，connected to the optical laser probe mock-up，are planned.If the improvement of the SNR will be conf i rmed，the LIDAR prototype will be able to recognize multiples echoes having large difference in amplitudes.

［1］Sabatini R，Richardson MA，Roviaro E.Development and f l ight test of an avionics lidar for helicopter and UAV low-level f l ight.J Aeronaut Aerosp Eng 2013；2:114.doi:10.4172/2168-9792.1000114.

［2］Pollastrone F， NeriC.Testresults fortriple-modulation radar electronics with improved range disambiguation.Fusion Eng Des 2015；96-97:912-16.doi:10.1016/j.fusengdes.2015.04.056.ISSN 0920-3796.

［3］Spinhirne JD.Micro pulse lidar.IEEE Trans Geosci Remote Sens 1993；31（1）:48-55.doi:10.1109/36.210443.

［4］Allen C，Gogineni S.A f i ber-optic-based 1550-nm laser radar altimeter with RF pulse compression.In:Geoscience and Remote Sensing Symposium，1999.IGARSS ‘99 Proceedings.IEEE 1999 International，vol.3.1999.p.1740-2，doi:10.1109/IGARSS.1999.772080.

［5］Allen C，Cobanoglu Y，Chong SK，Gogineni S.Performance of a 1319 nm laser radar using RF pulse compression.In:Geoscience and Remote Sensing Symposium，2001.IGARSS ‘01.IEEE 2001 International，vol.3. 2001.p.997-9，doi:10.1109/IGARSS.2001.976726.

［6］Gao S，Hui R.Frequency-modulated continuous-wave lidar using I/Q modulator for simplif i ed heterodyne detection.Opt Lett 2012；37:2022-4. doi:10.1364/OL.37.002022.

［7］Agishev R，Gross B，Moshary F，Gilerson A，Ahmed S.Range-resolved pulsed and CWFM lidars:potential capabilities comparison.Appl Phys B 2006；85:149-62.doi:10.1007/s00340-006-2254-6.

［8］CampbellJF，LinB，NehrirAR，HarrisonFW，OblandMD. Super-resolution technique for CW lidar using Fourier transform reordering and Richardson-Lucy deconvolution December 15，2014/Vol. 39，No.24/OPTICS LETTERS 6981.

［9］Cardarilli GC，Del Re A，Nannarelli A，Re M.Residue number system reconf i gurable datapath，Proc.IEEE International Symposium on Circuits and Systems，ISCAS 2002，II-756-II-759 vol.2，2002，doi:10.1109/ ISCAS.2002.1011463.

［10］Hamblen JO，Hall TS，Michael D.Rapid prototyping of digital systems. SOPC Edition Springer Science；2008.

［11］Campbell JF，Lin B，Nehrir AR.Advanced sine wave modulation of continuous wave laser system for atmospheric CO2 differential absorption measurements.Appl Opt 2014；53（5）.doi:10.1364/AO.53.000816.

Received 20 November 2015；revised 17 March 2016；accepted 13 April 2016 Available online 28 April 2016

Peer review under responsibility of China Ordnance Society.

*Corresponding author.Tel.:+39 06 9400 5535.

E-mail address:fabio.pollastrone@enea.it（F.POLLASTRONE）.

http://dx.doi.org/10.1016/j.dt.2016.04.002

2214-9147/? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.All rights reserved.

? 2016 China Ordnance Society.Production and hosting by Elsevier B.V.All rights reserved.