A MmWave Communication Testbed Based on IEEE 802.11ad with Scalable PtMP Configuration

Chaowei Wang,Mingliang Pang,Dinghui Zhong,Yuling Cui,Weidong Wang,2

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

2 Key Laboratory of Universal Wireless Communications,Ministry of Education,Beijing University of Posts and Telecommunications,Beijing 100876,China

Abstract:The sub-6G band is too crowded to accommodate higher data rate,while the millimeter wave(mmWave)bands have abundant spectrum resources and massive MIMO can provide high spectral and energy efficiency.Therefore,the combination of the two,namely mmWave-MIMO system,has attracted intensive research interests.In this paper,we develop a high-speed mmWave-MIMO communication system and conduct exhaustive field tests.The detail of the system design is provided and the key modules of the testbed are analyzed.The testbed exploits high gain of mmWave RF and flexible configuration of embedded system.The validation and field tests show that the developed testbed can provide up to 2.3 Gbps network layer data rate in single channel with low latency and support point-to-multi-point(PtMP)transmission aided by relay.The testbed can be used in future B5G and 6G systems to provide high reliability and low latency wireless coverage.

Keywords:mmWave testbed;point-to-multi-point;IEEE 802.11ad

I.INTRODUCTION

With the development of wireless networks,mobile data services are growing explosively.The International Telecommunication Union(ITU)has identified three major application scenarios for 5G: Enhanced Mobile Broadband(eMBB),Massive Machine Type Communications(mMTC),and Ultra-Reliable and Low-latency Communications(URLLC)[1,2].In order to meet the requirements of these application scenarios,future communication networks need to provide wider bandwidth,higher spectral efficiency,and accommodate much more users.One of the solutions to increase the system capacity for eMBB is improving spectral efficiency by physical layer technologies,such as Massive multiple-input-multiple-output and advanced channel coding technologies[3,4].In order to further improve local spectral efficiency,small base stations have been adopted[5].However,in existing cellular networks,the shortage of spectrum is the bottleneck of increasing capacity.Therefore,it is necessary to explore new frequency bands,including those not used in cellular communications.

Millimeter wave refers to RF signals with frequency between the microwave and light,with a frequency from 30 GHz to 300 GHz[6].Compared with the lower frequency band,the millimeter wave has much richer bandwidth resources,and is an important candidate frequency band for the future B5G and 6G networks to further improve the system capacity and even support the enhancement of Internet of Vehicles or other scenarios[7].In the past,millimeter wave communication was considered for outdoor point-topoint backhaul links[8]or indoor high-resolution multimedia streams[9].Currently,it has been proposed as a major candidate technique for the new spectrum of 5G cellular system and can build a communication system of data rate up to 10 Gbps with 2000 MHz super-bandwidth.There are other potential applications for mmWave as well.For example,with the recent excitement related to autonomous vehicles,mmWave may play an important role in providing high data rate connections among vehicles.This is coherent with the fact that mmWave is already the backbone of automotive radars.In the past ten years,automotive radars have been widely developed and deployed[10].The combination of mmWave communication and radar[11]is also interesting for mmWave applications.MmWave can be used to enable high-data-rate and low-latency connections to the cloud that permits remote driving of vehicles through novel mmWave vehicle-to-infrastructure.MmWave is also adopted in high-speed wearable networks that connect mobile phones,smart watches,augmented reality glasses,and virtual reality headsets[12].However,as the carrier frequency increases tens of times,the main obstacles to the mmWave are the high path loss and severe rain attenuation[13,14].

Recently,large-scale multiple-input–multipleoutput(MIMO)or “massive MIMO” systems have also drawn considerable interest from both academia and industry[15].Theoretically,massive MIMO systems can effectively relieve the inter-user interference in multiuser MIMO systems with simple linear transceivers[16,17].It is shown in[18]that each antenna element of a very large MIMO system consumes exceeding low power,and the total power can be made inversely proportional to the number of antennas.In addition,MIMO has many other advantages,such as high spectral efficiency,security,robustness and reliable linkage[18,19].Therefore,the massive MIMO also plays a key role in the next generation mobile networks.Besides,since the higher gain of the narrow beam generated by MIMO transmitter can compensate the propagation loss of high frequency signal[20],the mmWave-MIMO system can effectively alleviate the high path-loss and channel fading[21].Moreover,transmission of multiple data streams through spatial multiplexing can further improve the spectral efficiency[22].

Considering the above advantages,tremendous efforts have been devoted to the combined mmWave-MIMO system.It is demonstrated in[23,24]that large MIMO systems can enhance spectral efficiency by several orders of magnitude and simple zero-forcing transceivers are asymptotically optimal.Due to the necessity of channel state information(CSI)to realize massive MIMO performance gains,many different channel estimation algorithms have been designed for uplink channels[25].And downlink channels can be obtained by channel reciprocity for time division duplex(TDD)networks[26].With a huge number of antennas and the corresponding CSI,beamforming can reach a high spatial resolution[27].To reduce the cost of hardware implementation,hybrid analog-digital beamforming techniques have been developed for mmWave-band massive MIMO systems[28].In addition to the theories and algorithms,many mmWave-MIMO testbeds have been designed.[29]develops a mmWave testbed using commercial offthe-shelf devices and open-source software package,which uses a phase noise cancellation scheme to reduce the phase noise at the receiver.[30]presents a highly-integrated and reconfigurable MIMO-SAR testbed,along with examples of three-dimensional image reconstruction algorithms optimized for MIMOSAR configurations.Adopting a hybrid beamforming architecture and building upon software-defined radio technology,[31]designs an energy-efficient mmWave-MIMO communication system.

In this paper,we present a mmWave-MIMO communication testbed based on IEEE802.11ad,which has two configurations of point-to-point transmission(PtP)and point-to-multi-point transmission(PtMP).IEEE 802.11ad is also known as WiGig together with IEEE 802.11ay.The main contributions of this study are as follows:

1.As the future wireless networks require greater bandwidth,higher spectral efficiency,and accommodate much more users,we develop a mmWave-MIMO testbed combining mmWave and MIMO,both of which are key technologies in 5G.The testbed is easy to use,flexible to configure and can be used for indoor and outdoor wireless transmission.

2.Based on the mmWave and MIMO technology,we analyze the link propagation model,channel model and beamforming of the system.Then we give the received signal model of the testbed.

3.We test the indoor/outdoor performance of the testbed in various settings.The results show that when working with PtP configuration,the testbed can reach a peak rate of 2.21 Gbps with an average round-trip time(RTT)of 2.10 ms.The PtMP configuration of the testbed has a sum data rate of up to 2.3 Gbps with an average round-trip time of 2.75 ms.

The rest of this paper is organized as follows.Section II analyzes the key models of the testbed.Section III provides the system design and architecture of the developed mmWave-MIMO testbed.Section IV discusses the measurements and results verified by the testbed.Finally,the conclusion and a vision of future work are given.

II.THEORETICAL MODELS

In the World Radiocommunication Conference 2019(WRC-19),the globally unified millimeter wave frequency bands for International Mobile Telecommunications(IMT)were finally determined,e.g.,24.25-27.5 GHz,37-43.5 GHz,45.5-47 GHz,47.2-48.2 GHz and 66-71 GHz[32].According to IEEE 802.11ad,we choose the unallocated frequency band near 60 GHz as the transmission channel of our testbed.There are four candidate channels centered at:58.32 GHz,60.48 GHz,62.64 GHz and 64.80 GHz,and each channel has 2 GHz bandwidth.

As an important atmospheric attenuation window for mmWave,V-band spectrum provides adequate frequency channels with much less interference,which is ideal for high data rate radio transmission.Even though,V-band also has shortages such as high path loss and large penetration loss[33].In order to overcome the restrictions,the transmitting antennas exploit the high gain of the directional narrow beam,which also facilitate the information security.A high data rate can be provided within the coverage area of the system while signals beyond this range attenuate dramatically.Conclusively,the testbed is especially suitable for indoor/outdoor LOS(line of sight)transmission without blocking.

In this section,we analyze the key models of mmWave communication system to provide a theoretical basis for our testbed.

2.1 Link Propagation Model

Propagation is unique at mmWave due to the small wavelength compared to the size of most of the objects in the environment.Analyzing these channel characteristics is fundamental to developing signal processing algorithms for mmWave transceivers.

We adopt the WiGig link propagation model in[34].Based on this model,when the distance between two transceivers isr,the received signal power at Rx end can be expressed as:

whereGo[dB]is the maximum achievable gain of the antenna andθTXis the TX beam center angle.θ-3dBis its half power beam width.

2.2 Spatial Characteristics and Multipath Channel Models

The mmWave MIMO channel can be described with standard multipath models used in lower frequencies[36].We consider a MIMO system withNTtransmitting antennas andNRreceiving antennas.For 2D channel model,the transmitting and receiving antenna arrays are described by their array steering vectors aT(θT)and aR(θR),which represent the phase profile of the array as a function of the angular directionsθRandθTof arriving or departing from the plane wave[37].For an N-element uniform linear array(ULA),the spatial steering vector can be expressed as

Obviously,the equations(9),(10)and(11)show the channel model influenced by the dual-wideband effects.And the channel in equations(9)is named as the spatial frequency wideband(SFW)channel,which provides a more accurate channel model for largescale antennas.

2.3 Beamforming for V-band Transmission

In order to reduce the inter-user interference of the system,we adopt the beamforming in our testbed.Beamforming is widely used in interference management.In[39],the author proposes an interference alignment and soft-space-reuse(IA-SSR)-based cooperative transmission scheme,in which different algorithms are applied for the cell-edge and cell-center clusters.To deal with the unfair service problem,the author introduces the IA method based on two-stage precoding framework for cell-edge users to enhance their throughput as

2.4 Received Signal Model

wherenis the Gaussian noise added to the received signal,which has a mean of 0 and a variance ofσ2.

III.SYSTEM DESIGN AND ARCHITECTURE

In this section,we demonstrate the architecture of the system and the design of the antenna and transceiver in detail.

3.1 Antenna Design

The antenna array is based on the antenna that we presented in[43,44],which is a wideband circularly polarized(CP)microstrip antenna.Based on a square patch with slot-loading fed by dual-feed wideband feed network(WFN)and a frequency selective surfaces(FSS)structure,the CP and polarization diversity are achieved,and antenna array coupling is greatly decreased.

The WFN consists of a coupled-line Wilkinson power divider and a stub-loading 90º phase shifter,which are illustrated in Figure 1.As shown in Figure 1,the input signal to port 1 is divided into two sections with the same amplitude after going through the power divider.Then the 90ºphase difference is obtained after passing the phase shifter without changing the amplitude.At the output of the WFN,two ways of the orthogonal signal at port 2 and port 3 can be achieved.

Figure 1.Structure of WFN.

Figure 2.Geometry of the proposed MIMO antenna.

Figure 3.Photograph of the fabricated antenna.

The designed antenna structure with two antenna elements is demonstrated in Figure 2.The square shaped radiating patches with quad-symmetrical bent slots loading are placed on the top layer of the middle FR4 substrate.The square shaped patches are adopted because they can make sure that the two ways of signal provided by the WFN are kept orthogonal and then perfect CP radiation can be excited.In addition,polarization diversity among the antenna array elements can easily be realized simply by taking mirror inversion of the signal feeding port 3.According to Figure 2,each one of the patches generates left-hand CP(LHCP)waves and the right-hand CP(RHCP)waves respectively.

Figure 4.The system architecture of the testbed.

The WFNs are placed on the bottom layer of the bottom FR4 substrate,and they feed the patches through four probes,while the ground plane is placed on the top layer.The square patches with square-ring slot loading are placed on the top layer of the upper FR4 substrate to form the frequency selective surfaces(FSS)structure,which can be used to enhance the antenna gain[45].The FSSs are placed above the square radiating patches.

Based on the above design,our antenna has the characteristics of low loss and large bandwidth.At the same time,as the beam is narrow,our antenna has a high gain suitable for mmWave communication.And the prototype of the antenna is shown in Figure 3.

3.2 System Design

Figure 4 demonstrates the system architecture of the developed testbed.The mmWave transceiver,with the prototype shown in Figure 5,consists of three parts:the RF module,the baseband module and the networking board based on the embedded system.The transceiver works in TDD mode.When working as a transmitter,it has three tasks: a)getting data from the UE and processing these data signals in the baseband;b)converting the baseband signals into mmWave signals;c)transmitting the mmWave signals to the receiver in V-band.Correspondingly,when it works as a receiver,its tasks are: a)receiving mmWave signals from the transmitter;b)converting mmWave signals to baseband signals;c)reconstructing the original information in the baseband signals and passing it to the UE.

Figure 5.The transceiver of mmWave-MIMO testbed.

Figure 6.The architecture of PtP configuration.

Figure 7.PtMP configuration of mmWave-MIMO testbed.

The ARM-based networking board is connected to the user equipment(UE)with RJ-45 interface.We implement the design of embedded system on this module,which allows the UE to directly transmit data with the transceiver.The RF module and baseband module are connected to the networking board through USB3.0 port.The networking board acts as a switch or transparent bridge in the system.

We adopt the SoC PRS4601[46],which is manufactured by Peraso,to support 802.11ad PHY and MAC protocol.The PRS4601 PHY is capable of modulating/demodulating withπ/2-BPSK,π/2-QPSK and 16-QAM and achieves up to a maximum PHY layer rate of 4.62 Gbps.The SoC uses AES with a 128-bit key to ensure the security.

The RF module is composed of upconverter,downconverter and 2-element antenna array.The antenna array adopts the design introduced in section 3.1.It has a Tx output power of 15 dBm and Rx noise figure of 6 dB.

IV.FIELD TEST RESULTS

In this section,we first present the testbed configuration.Then we validate the performance of the developed testbed by testing the antenna performance,system’s data rate,and round-trip time.

4.1 Testbed Configurations

According to the system architecture we analyzed in Section III,the transceiver can be configured in three modes: base,relay and client.In base mode,the function is to act as a base station and provide wireless coverage.The relay mode means relaying signals between the base station and the users.The relay can increase the wireless coverage range and enhance the link performance by increasing the system throughput.In client mode,the transceiver acts as a user equipment.

Based on the above three modes,there are two communication configurations: point-to-point transmission(PtP)and point-to-multi-point transmission(PtMP).Figure 6 shows the architecture of the PtP configuration and Figure 7 shows the transceivers’configuration in PtMP.The AP in Figure 7 is a transceiver set to base station mode,the Relay is a transceiver set to relay mode and the UE is a transceiver set to client mode or a computer.The PtMP configuration is assisted by relay.The main parameters of our testbed are listed in Table 1.

Table 1.Parameters of mmWave-MIMO testbed.

Figure 8.The antenna measurement and test setup.

4.2 Test Environment

Before describing the test environment,we first introduce the UE.It can be a computer or any device capable of communication.It is connected to the transceiver via Ethernet cable(RJ-45).As the source of system traffic,UE generates data packets and passes them to the transceiver(or receives data packets by the receiver).By transmitting and receiving the generated data packets,the UE could evaluate data rate,roundtrip time(RTT)and other performance indicators.

The validations consist of the antenna measurement and the system performance test.The measurement of the antenna patterns is carried out in an anechoic chamber shown in Figure 8.

With UEs and the transceivers,our system test environment is divided into indoor and outdoor scenarios shown in Figure 9.The one on the left shows the indoor test layout.Due to the limited space,the transmitting distance is no more than 10 meters.The one on the right in the figure shows the outdoor test environment,the outdoor test is carried out in the campus arena.The maximum dimension of the playground is more than 200 meters,and the ground is flat without obstacles.

Figure 9.The test environment of the system.

Figure 10.The constellation of the received signal.

4.3 Constellation of the Received Signal

In order to further verify our received signal model,we observe the received signal constellation.The constellation is regarded as a “two-dimensional eye diagram” of digital signals,which has an intuitive effect on judging the bit error rate of the received signal.The received signal constellation of the testbed under different SNRs is shown in Figure 10.It can be observed that the constellation points get distributed when SNR is lower.

Figure 11.Simulated and measured radiation patterns of the antenna.

Figure 12.The EIRP of antenna.

4.4 Antenna Mearsurement

In this section,we measure the designed antenna of the testbed,such as the antenna radiation pattern and Effective Isotropic Radiated Power(EIRP).

The antenna radiation pattern,which is one of the most important tools for analyzing the antenna,refers to the pattern of the relative field intensity(normalized modulus)of the radiation field changing with the direction at a certain distance from the antenna.It is a graphical description method of the antenna radiation characteristics.The antenna parameters can be observed from the antenna pattern.The results of radiation patterns at 60 GHz in the two principal planes are illustrated in Figure 11.The isolation between LHCP and RHCP is higher than 20 dB at the boresight direction.

EIRP is usually used to measure the intensity of interference and the ability of a transmitter to send signal.We measure the EIRP of the antenna in channel 2,which centered at 60.48 GHz in Figure 12.The test method is from-90 degree to +90 degree,with 1 degree as a step for scanning.As shown in Figure 12,the maximum transmitting power of the antenna is around 34 dBm,which is slightly lower than the 35 dBm we designed.And the fluctuation of antenna is also within the normal range.

Figure 13.The indoor variation of data rate with distance.

Figure 14.The outdoor variation of data rate of with distance in PtP configuration.

Figure 15.The indoor variation of data rate with angle in PtP configuration.

4.5 Data Rate

Data rate is one of the key indicators of the communication system performance.We test the variation of average data rate and peak rate with distance in the two scenarios,as well as the relationship between average data rates and angles indoor.

Figure 13 shows the indoor variation of data rates with distance in PtP and PtMP configuration of the testbed respectively.In PtMP configuration,we calculate the sum rates of the two UEs.As shown in Figure 13,the testbed indoor can achieve an average data rate of 2 Gbps.The last two rates of both scenarios have significant decrease as our test room is too small and the walls bring in serious reflection and reduce the data rate.The peak rate in both figures fluctuates dramatically due to dense scattering of the signal in the small test room.

The data rates with outdoor longer distances are shown in Figure 14.With a distance shorter than 100 m,our testbed can still provide a stable data rate of 2 Gbps.When the distance increases to 150 meters,the data rate drops to 1.816 Gbps,which is 9.3% lower than the rate at 50 m.As the distance continues to increase,the data rate drops significantly.When the distance reaches 200 m,the testbed failed to provide a stable connection.That is because the transmission power is limited and the transmission height is not high enough.

The variation of data rate with angle in PtP configuration is tested indoor,the result is shown in Figure 15.Obviously,our testbed can provide a stable connection in the angle of 10 degree to-10 degree.As the angle increases,the data rate decreases rapidly.

Figure 16.The RTT of testbed.

In general,our testbed provides high-speed and reliable connectivity in both indoor and outdoor scenarios.It can provide a data rate of around 2 Gbps in both indoors and outdoors scenarios within 150 m without obstacles,which is difficult for existing wireless transmission techniques.

4.6 Round Trip Time

Round-trip time(RTT)is a packet’s two-way travel time between the server and UE,which includes packet-propagation time,packet-queuing time and packet-processing time[47].The RTT reflects the network congestion to a certain extent.It is also an indicator of transmission delay in network layer.

We test the RTT with the two configurations indoor as shown in Figure 16.We ping 1000 times with each configuration,and the packet size is 64000 bytes in each time.When the testbed works with PtP configuration,the average round-trip time is 2.10 ms,which is generally from 6.5 ms to 20.5 ms in LTE.When the testbed works with PtMP configuration,the RTT is 2.75 ms.It is slightly higher than the PtP configuration due to the increased processing delay.And the packet loss rates in both configurations are 0 when link is established.This further proves that the connection provided by our testbed is stable and reliable.

V.CONCLUSION

In this paper,we design a mmWave-MIMO communication structure supporting PtP and PtMP configurations.We model the mmWave link propagation,multipath channel model and received signal.The implementations of beamforming,networking and I/O interface are provided.Then we develop the transceiver complying with IEEE 802.11ad,it can be deployed indoor and outdoor.The testbed performance is evaluated by field tests,and the results show that the testbed provides reliable two-way communication with a steady data rate of 2 Gbps and very low latency for indoor and outdoor scenarios.In future work,we will aggregate the 4 channels with link aggregation to achieve even much higher data rate of 10 Gbps.

ACKNOWLEDGEMENT

This work was supported by National Key R&D Program of China(2020YFB1807204).