A Satellite Communication System Transmission Scheme Based on Probabilistic Shaping

Wei Zhang,Xia Sheng,Xishuo Wang,Qi Zhang

1 Science and Technology on Information Systems Engineering Laboratory,National University of Defense Technology,Fuyuan road NO.1,Changsha,410000,China

2 China Academy of Space Technology (CAST),Beijing 100094,China

3 School of Electronic Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China

4 Beijing Key Laboratory of Space-Ground Interconnection and Convergence,Beijing University of Posts and Telecommunications,Beijing 100876,China

5 State Key Laboratory of Information Photonics and Optical Communications,Beijing University of Posts and Telecommunications,Beijing 100876,China

Abstract: In this paper,A transmission scheme based on probabilistic shaping applied to satellite communication systems is proposed.16QAM is taken as an example to establish a 1GBaud ROF experimental system working in Ka-band.The experiment results show that the PS-16QAM signal has better performance in terms of bit error rate than the uniform 16QAM,and its performance is close to the uniform 8QAM scheme.

Keywords: probabilistic shaping; satellite communication; Ka-band

I.INTRODUCTION

In today's rapid development of communication and network,timely information interaction has become an essential basic requirement for people and promoted the development of terrestrial networks.However,in many special scenarios,such as sea,air and remote areas,the ground network cannot cover and the cost of equipment is laid,so that the traditional terrestrial network cannot meet the communication needs.Satellite communication technology has been closely watched by all parties[1-5].With the rapid growth of the demand for satellite broadband multimedia services,the C and Ku band resources have been very limited and cannot meet the requirements of today's communication quality and capacity.The Ka-band plays an important role in the development of satellite communications in the 21st century due to its high available bandwidth and low interference.The frequency band provided by the Ka band is 3.5 GHz,which is 5 to 8 times the bandwidth of the Ku band,and has available bandwidth (27.5 to 31 GHz in the third region,17.1 to 21.2 GHz in the downlink,3500 MHz in the wideband),large transmission capacity,and less interference,the equipment is small and so on[6-8].The advantages of small equipment and small size play an increasingly important role in two-way multimedia satellite services and high-speed satellite communications,especially for high-speed satellite communications,broadband digital transmission,HDTV,SNG,VSAT,DTH television and other two-way satellite multimedia[9-10].

In recent years,more countries in the world have begun to develop or prepare to develop Ka-band communication services.The Kaband satellite transponder uses a cellular multi-spot beam similar to the terrestrial network.It can have a higher equivalent isotropic radiated power (EIRP)and obtain higher power under the same size antenna,so the Kaband receiving antenna can be made smaller,easy to miniaturize and enhance portability.At the same time,in the case of using spot beams,the frequency multiplexing can be performed by the method of spatial isolation.The bandwidth is multiplied,and the cost of the Ka unit bandwidth is significantly reduced.In order to mitigate the impact of rainfall on the cognitive radio geostationary (GEO)satellite operating in the Ka-band,an adaptive modulation scheme for rainfall fading prediction using Kalman filtering is proposed [11].The authors of [12] studied the Universal Space Transponder-Ka-Band Modulator (UST-KaM)developed by NASA's Jet Propulsion Laboratory for the NASA-ISRO SAR (NISAR)mission,using OQPSK with baseband and RF filtering to include 1Gsps transmission spectrum in the 1.5GHz Ka band.In the existing transmission schemes,most of the conventional modulation formats are used,and high signal transmission power is often required and the spectrum efficiency of the system is low.

It is well known that for bandwidth-constrained additive white Gaussian (AWGN)channels,Shannon Limit cannot be achieved by using only uniform M-PAM (or M-QAM)constellations,and its performance loss tends to 1.53dB,the so-called shaping gain.In the Ka-band satellite communication system,in order to make the system capacity closer to the channel capacity,a high-order modulation can be used,but as the order increases,the average distance between the constellation points becomes smaller,and it is susceptible to weather.In the affected Ka-band,the received signal bit error rate will increase significantly,Moreover,the average power of the transmitted signal is also increased compared to the low-order modulated signal,resulting in unnecessary power waste.In addition,the complexity of the modem device is high,which is difficult in practical applications.Therefore,in order to reduce the adverse effects of weather on communication quality in Ka-band satellite communication systems,obtain high signal-tonoise ratio gain and spectral efficiency,channel coding technology must be combined with constellation shaping.The constellation shaping increases the probability of occurrence of lower energy signal points in the middle of the constellation by increasing the distribution of constellation points,and reduces the probability of occurrence of constellation points with higher energy,thereby reducing the transmission power of the system.Besides,probabilistic shaping has an advantage of increasing the spectral efficiency of the Ka-band satellite transmission system furtherly.[13-16]As far as we know,there has been no report on the introduction of probabilistic shaping into the Ka-band satellite communication system,so the main contribution of this paper is as follows.

In this paper,a probabilistic shaping transmission scheme applied to satellite communication systems working on the Ka-band is proposed.And in order to verify the performance of our proposed solution,a 1GBaud ROF experimental system is built.This paper is divided into the following sections.Firstly,the optimal signal distribution according to channel characteristics is introduced.The second part introduces the transmission process based on probabilistic shaping technology.Finally,it proves the proposed by establishing a radio-over-fiber(ROF)system with 1GBaud.

II.PRINCIPLE

The satellite communication transmission system model operating in the Ka-band is shown in figure 1.The probabilistic shaping transmission scheme applied to the Ka-band proposed in this paper is shown in Figure 2.The bit sequence to be transmitted passes through the distribution matcher and then an amplitude sequence obeying the channel characteristics is outputted.The amplitude sequence is encoded after binary labeling by systematic LDPC and then modulated it using M-QAM.Finally,the sequence is sent into the satellite channel for transmission.Satellite received signal is transmitted to ground receiver through satellite-to-ground channel.QAM demodulation,LDPC decoding,inverse binary labeling and inverse distribution matcher are performed at the receiving end.The specific process is described below.

Fig.1.Satellite communication transmission system model.

Fig.2.Satellite communication probabilistic shaping transmission system.

2.1 Optimal signal input

It is well known that changing the uniformly distributed constellation into an optimal distribution can reduce the power of the transmitted signal effectively,thereby obtaining the shaping gain.

Let the probability of the constellation point λ represented asP(λ).For a constellation with a fixed average energy limitthe optimal probability distribution for each constellation point satisfies equation (1)

where the parameter η is non-negative used to make a trade-off between average energyand bit rate α.In practice,the probability distribution of the input signal is not continuous.For discrete inputs,the probability of each constellation point is expressed by equation (2)

where the parameterληensures that the sumPλη(λ)is 1.Pλη(λ)distributes around 0 symmetrically which is a well-known distribution in statistical mechanics and thermodynamics - the Maxwell-Boltzmann distribution[17].

2.2 Transmission system based on probabilistic shaping

Complex modulation formats such as quadrature amplitude modulation (QAM)are used as the most common modulation format for increasing data rates for today's communication systems.This paper takes 16QAM as an example to study satellite transmission system.16QAM is considered to be the most promising modulation method for single-carrier 400G transmission.It can be seen from equation (2)that it is symmetric around 0.Based on this,a signal can be written as the product of its amplitude and sign.Since the amplitude and sign of the signal are two independent variables,the probability of one signal point is the product of the amplitude probability and the sign probability.The value of the signal sign can only take 1 or -1,so the probability of the symbol is 1/2,which is a constant.In order to make the distribution of the signal point obey the Maxwell-Boltzmann distribution,it is only necessary to ensure that the amplitude distribution of the signal is optimally distributed.

This scheme adopts the above probabilistic amplitude shaping(PAS)[18],and proposes a transmission scheme of satellite communication system based on probabilistic amplitude shaping.The specific process is as follows.First,the bit data sequence to be transmitted is generated by the distribution matcher to generate an amplitude sequence satisfying the optimal distribution,Maxwell-Boltzmann distribution.The distribution matcher uses constant composition distribution matching(CCDM)[19].ncis supposed the length of the amplitude sequence output of the distribution matcher.The probability of a certain amplitude is the product of the number of occurrences of the amplitude in the ncamplitude sequence and 1/nc.Through the algorithm 2 in [20],the number of occurrences of each amplitude of the amplitude sequence in accordance with the optimal distribution can be obtained.Since the position at which each amplitude occurs is unknown,it is necessary to consider all kinds of possibilities,all of which may be represented by a binary sequence,each binary sequence representing the arrangement of the amplitudes in an amplitude sequence.Matching the data bits to be transmitted with all possible amplitude sequences to achieve an optimal amplitude distribution sequence and complete the function of distribution matching.

The resulting optimally distributed amplitude sequence is converted to a binary sequence subject to the optimal distribution after passing through the binary label.In the transmission process,due to the influence of light refraction and optical signal fading,bit error occurs during the process of reception.In order to avoid bit error,LDPC is used as the error correction coding of the channel,but to prevent the sequence before and after the coding.The optimal distribution characteristics are corrupted,so systematic LDPC coding is required.After the systematic LDPC coding,the information sequence can be retained,except that the modulo 2 of the data bits is added after the information sequence as the coding redundancy,so that the original distribution is preserved.

The encoded bit sequence is mapped into the 16QAM constellation to complete the modulation of 16QAM.The transmission process is then completed through the Ka-band satellite channel model.Because weather factors have a great influence on Ka-band electromagnetic wave,in the Ka-band geostationary orbit communication satellite system,weather fading is the most important factor affecting the transmission quality of the system.Therefore,the channel model we used refers to the data obtained by C.Loo's propagation characteristics measurement by Olympus star[21].Whereμ1andσ1are the measurement parameters of the envelope in different weathers,μ2andσ2are measurement parameters of the measured parameters of the phase in different weathers.The channel model established based on these four parameters is as shown in equation (3).The receiving process is the reverse process of the transmitting process,and will not be described here.

The upper bound of the symbol error rate (SER)of the received signal can be expressed by equation (4)[22],wheresiandsjare possible symbols in theN-dimensional constellation,their bits mappings are represented byβi,N0is theN-dimensional Gaussian noise variance,andi s the norm of vectors.The Hamming distance of each symbol pair needs to be considered when converting SER to bit error rate (BER)h(β i,βj),as equation (5)shown.

III.EXPERIMENT SETUP

The experimental setup for the probabilistic shaping Ka-band satellite communication system is showed in figure 3.We adopted a radio-over-fiber system to obtain the millimeter-wave wireless signal by heterodyne beating.At the transmitter side,a continuous-wave (CW)light wave is generated from an external cavity laser (ECL1)at 193.1080 THz with a linewidth less than 100kHz,which is applied as a signal source and fed into an in-phase/quadrature (I/Q)modulator (MOD)with 33GHz bandwidth and 6dB insertion loss.The baseband signal is generated with PRBS by different constellation mapping,including well-studied modulation formats such as QPSK,8QAM and 16QAM,as well as the novel probabilistic shaped 16QAM(PS-16QAM).Arbitrary waveform generator (Tektronix AWG70002)is utilized to generate 1GBaud I/Q electrical baseband signals.The driving signal's amplitude is boosted to 2Vpp by an electrical amplifier (EA)to drive the IQ MOD biased at the linear regime.An Erbium Doped Fiber Amplifier (EDFA)is used to compensate the power loss during modulating.Then we use an optical coupler (OC)to combine the modulated optical signal with another light wave generated from ECL2 at 193.1110 THz,which works as a local oscillator (LO),to produce the frequency offset of 30 GHz between LO and received optical signal.An attenuator is employed to control the power of the received optical.Then the 30GHz radio frequency(RF)signal is generated by the beating of the two light waves at the PD (3dB bandwidth of 70GHz and responsivity of 0.6A/W).After the amplification by EA,the generated Ka-band signal is sent into free air by a horn antenna with 20dBi gain.At the receiver side,the Ka-band wireless signal is received by a matched antenna.After boosted by the same EA as the first one,the IF signals (IF = 30 GHz)is converted into digital signals at the sample rate of 100 GSa/s by the oscilloscope with 20 GHz bandwidth.Finally,the recorded digital signal is processed by offline digital signal progressing (DSP)including down conversion,down sampling,CMA equalizing,phase recovering and demapping as shown in figure 4.

The constellation received at the receiving end is as shown in figure5.We compared the bit error rate performance of uniform 16QAM,PS-16QAM and 8QAM in the same experimental system.When the baud rate is 1GBaud and the received optical power is different,the error rate of each scheme is as shown in the figure6.From the experimental results,it can be seen that PS-16QAM scheme gets a great improvement over the performance of the uniform 16QAM scheme.At the same time,PS-16QAM scheme is not only more efficient than the uniform 8QAM scheme,but also obtaining the error performance close to the uniform 8QAM scheme.

Fig.3.ROF Experimental system of Ka-band.

Fig.4.Procession of the offline digital signal progressing.

Fig.5.Constellation of uniform -16QAM,PS-16QAM,uniform-8QAM.

IV.CONCLUSION

In this paper,a probabilistic shaping 16QAM scheme for the Ka-band is proposed,which is applied to the satellite communication system.Besides,a 1GBaud ROF 16QAM experimental system is built to validate the scheme.The modulated Ka-band signal is sent to 2.5 meters atmospheric wireless link.The experiment compares the error performance of uniform 16QAM,PS16QAM and 8QAM.The experimental results show that the PS-16QAM signal have better performance in terms of bit error rate than the uniform 16QAM signal,and its performance is close to the uniform 8QAM scheme.A high-performance transmission scheme,used for the use of satellite communications in the millimeter-wave band,is proposed and fully validated.

Fig.6.BER versus received optical power.

ACKNOWLEDGEMENT

This work is partly supported by National Nature Science Foundation of China (NSFC)(61675033)