6G sensing and communication convergence · 2021. 4. 9. · Petteri Kela, Kari Leppanen, and Mikko Valkama. "High-efficiency device positioning and location-aware communications in

Chalmers University of Technology

6G sensing and communication convergence

Henk Wymeersch and Tommy SvenssonDepartment of Electrical EngineeringChalmers University of TechnologyGothenburg, Swedenemail: {henkw,tommy.svensson}@chalmers.se

© Henk Wymeersch and Tommy Svensson, 2020 1

6G Channel: 6G Research Visions Webinar SeriesLocalization and Sensing – Technologies, Opportunities and ChallengesNov 18, 2020


Outline

• Why 6G needs localization and sensing• 6G as a sensor for localization and sensing (bistatic view)• Communication and sensing convergence (monostatic view)• Location-aided communication• What is next?



Why 6G needs localization and sensing

Status in 5G mmWave• Beam alignment and beam tracking• Blockage avoidance by beam search• Proactive resource allocation• 5G can benefit from location and map information

Vision in 6G• Finer beams: even more overhead• Environment maps for discovery of propagation paths• 3D orientation and 3D location information needed• New sensor: a communicating 6G radar• 6G must harness location and map information


Bui, Nicola, Matteo Cesana, S. Amir Hosseini, Qi Liao, Ilaria Malanchini, and Joerg Widmer. "A survey of anticipatory mobile networking: Context-based classification, prediction methodologies, and optimization techniques." IEEE Communications Surveys & Tutorials 19, no. 3 (2017): 1790-1821.

Giordani, Marco, Marco Mezzavilla, and Michele Zorzi. "Initial access in 5G mmWave cellular networks." IEEE Communications Magazine 54, no. 11 (2016): 40-47.

Koivisto, Mike, Aki Hakkarainen, Mario Costa, Petteri Kela, Kari Leppanen, and Mikko Valkama. "High-efficiency device positioning and location-aware communications in dense 5G networks." IEEE Communications Magazine 55, no. 8 (2017): 188-195.

Tariq, Faisal, Muhammad RA Khandaker, Kai-Kit Wong, Muhammad A. Imran, Mehdi Bennis, and Merouane Debbah. "A speculative study on 6G." IEEE Wireless Communications 27, no. 4 (2020): 118-125.

https://developer.apple.com/

https://developer.apple.com/


6G as a sensor for localization and sensing

Status in 5G mmW• Localization with TDOA, AOD, RTT• Multipath: foe to friend• 5G as bistatic radar: allows mapping the environment (SLAM)

and tracking passive objects (SLAT)• Ability to share location, map, object information

Vision for 6G1. High carrier frequencies (above 0.1 THz): fewer paths2. Large bandwidths (above 1 GHz): better delay resolution3. Large number of antennas (> 100 for SNR gain): better angle

resolution4. D2D communication: relative positioning, bistatic radar 5. Network densification: short LOS links6. Intelligent metasurfaces: shaping the environment for improved

positioning quality 7. AI/ML: solve hard problems with data-driven approach


Base Station (BS)

User Equipment (UE)

From the signal sent by a single basestation in an unknown propagationenvironment, can we determine the UEposition, heading, clock bias?

Even when the line-of-sight is blocked?

And provide radar-like mapping capabilities?

And shape the propagation environment?

“3GPP TR 38.855 V16.0.0; study on nr positioning support,” Tech. Rep., 2019.

Witrisal, Klaus, Paul Meissner, Erik Leitinger, Yuan Shen, Carl Gustafson, Fredrik Tufvesson, Katsuyuki Haneda et al. "High-accuracy localization for assisted living: 5G systems will turn multipath channels from foe to friend." IEEE Signal Processing Magazine 33, no. 2 (2016): 59-70.

Ge, Yu, Fuxi Wen, Hyowon Kim, Meifang Zhu, Fan Jiang, Sunwoo Kim, Lennart Svensson, and Henk Wymeersch. "5G SLAM Using the Clustering and Assignment Approach with Diffuse Multipath." Sensors 20, no. 16 (2020): 4656.


6G as a sensor for localization and sensing

• Channel charting and AI– When channel is described by too many parameters– Data-driven approach and dimensionality reduction– Generates a pseudo-map that can be used as an

unsupervised replacement for radio maps

• Intelligent surfaces– Multipath: from foe to friend to partner– New reference points– Can provide automatic synchronization and AoD value– Wavefront curvature and near-field propagation effects


Radio-Frequency Propagation

In real life, geometric propagation models are intractable.The set of parameters explodes

Too many of them are unknown and/or impossible to estimate(e.g. scatterers at unknown position or with unknownelectromagnetic properties)

This suggests a data-driven approach: Channel Charting [1]

[1] C. Studer, S. Medjkouh, E. Gönültaş, T. Goldstein, and O. Tirkkonen, “Channel Charting: Locating Userswithin the Radio Environment using Channel State Information,” IEEE Access, 2018

Examples of Channel Charting Results

Channel charts (d = 2) obtained on the same training dataset.Zhang, Chuan, Yeong-Luh Ueng, Christoph Studer, and Andreas Burg. "Artificial Intelligence for 5G and Beyond 5G: Implementations, Algorithms, and Optimizations." IEEE Journal on Emerging and Selected Topics in Circuits and Systems 10, no. 2 (2020): 149-163. 2

Figure 2. Application examples of RIS-based localization and mapping services (from left to right): (i) LoS blockages can be circumvented to improvelocalization accuracy and continuity; (ii) wavefront curvature in the near-field of a large RIS receiver or transmitter can be exploited to solve for nuisanceparameters (e.g., clock biases); (iii) by creating strong and consistent multipath, RISs can support localization in very harsh indoor environment, dynamicallyaccounting for object movements; (iv) new delay-sensitive, ultra-accurate applications will be supported by the fact that RISs do not introduce processingdelays.

through technology whereby surfaces are endowed with thecapability to actively modify the impinging electromagneticwave [2], as visualized in Figure 1. A RIS can be implementedusing a variety of technologies as discussed below and canprovide significant benefits in terms of communication byguaranteeing coverage when the LoS is blocked. A RIS canoperate as a reconfigurable mirror or as a reconfigurable lens(see Figure 1). The RIS is controlled by a local control unitthat adjusts the phase profile or current distribution. Basedon these fundamental operating modes, a RIS can act as atransmitter [2], receiver [3], or as an anomalous reflector,where the direction of the reflected wave is no longer specularaccording to natural reflection laws but steerable [4], [5]. TheRIS concept can be applied at different wavelengths, rangingfrom low sub-6GHz bands, where the technology is wellunderstood and commercial systems are available, to 28 GHzmmWave bands, where RISs can provide significant benefitsin terms of coverage but where the technology is less mature.Finally, in the 0.1–1 THz regime, severe path loss, highersusceptibility to blockages, atmospheric absorption, and rainattenuation as well as significant hardware limitations makeRIS design challenging but can also lead to large performancegains.

The aforementioned properties and their close relation to theenvironment geometry make RIS attractive for localization andmapping. The potential of RIS for localization has receivedonly limited coverage in the literature, including preliminarystudies where the RIS operates in receive mode as a lens [6]and in reflection mode [7]. Hence, it is timely to delve deeperinto the potential of RIS for localization and mapping, as wellas the main research questions that we should address in thecoming years. Possible applications of RIS for localization arevisualized in Figure 2.

This paper aims to take a broader view than the technical

contributions in [6], [7] by describing the core technicalchallenges of applying RISs to localization and mapping,along with a preliminary system vision, results, and solutionsrecently put forward on related topics.

RADIO LOCALIZATION AND MAPPING

Basic Principles

Any radio localization and mapping system comprises threeessential parts: measurements, a reference system, and theinference algorithms.

Measurements: The measurements are derived from theradio signal between a transmitter and a receiver. They cantypically be obtained directly from the channel estimationroutine used for communication. Common location-dependentmetrics are based on received signal strength, time of arrival(ToA), phase of arrival (PoA), angle of arrival (AoA), angle ofdeparture (AoD), and Doppler measurements. Measurementscan be characterized by their resolution and accuracy. Theresolution refers to the ability to distinguish two signals basedon their measurements and depends on the signal bandwidthand duration, carrier frequency, the number of antennas, andcoherent integration time. The accuracy refers to the extent towhich we can determine the parameter of interest. It dependsalso on the signal-to-noise ratio (SNR), as well as on thedetailed properties of the signal waveform such as the time-frequency and spatial power allocation.

Reference System: All the measurements are taken in acertain frame of reference, e.g., that of the receiver. Refer-ences, sometimes called anchor points, have known states.There may be multiple position references, as in cellularlocalization or satellite positioning, which may in turn placerequirements in terms of synchronization, array calibration, aswell as dedicated control signals. The geometric placementof the reference plays an important role in the accuracy of a

Wymeersch, Henk, Jiguang He, Benoît Denis, Antonio Clemente, and Markku Juntti. "Radio localization and mapping with reconfigurable intelligent surfaces." arXiv preprint arXiv:1912.09401 (2019).

scatter point

-2 0 2 4 6 8

0

5

10

3

3.5

4

4.5

reflector

-2 0 2 4 6 8

0

5

10

0.4

0.6

0.8

1

1.2

RIS

-2 0 2 4 6 8

0

5

10

1

2

3

4

Figure 4. PEB as a function of the user location for a scatter point, a surface,and the optimized RIS with K = 1. W = 100 MHz.

scatter point

-2 0 2 4 6 8

0

5

10

1

2

3

4

reflector

-2 0 2 4 6 8

0

5

10

0.2

0.4

0.6

0.8

1

1.21 RIS

-2 0 2 4 6 8

0

5

10

1

2

3

4

Figure 5. PEB as a function of the user location for a scatter point, a surface,and the optimized RIS with K = 1. W = 1 GHz.

2) Spatial PEB: Fig. 4–5 show the spatial distribution ofthe PEB (capped at 5 m) for the scatter point, reflector and asingle RIS (K = 1). For the scatter point, the PEB takes onlow values close to the scatter point, but not so close that thepaths are no longer resolvable. For the reflector, the PEB is lowwhen the reflected path is available. For the optimized RIS,a PEB less than 5 m can be achieved throughout most of thedeployment region. When the bandwidth is larger, all scenariosbenefit from better delay resolution, in particular close to theobject. For all cases, we note that there are white regions wherethe PEB is worse than 5 m. We discern 2 cases: (i) whereone can not issue a unique positioning result (when only 1path can be resolved), so that PEB = +1, (ii) where thePEB exceeds the 5 m threshold, due to a poor diversity ofinformation directions and/or poor SNR (despite a sufficientnumber of resolved paths).

3) PEB CDF: In Fig. 6-7 we visualize the PEB results as acumulative density function (CDF). For 100 MHz bandwidth,the reflector offers sub-meter performance for about 45% ofthe deployment region. The RIS covers over 80 % with PEBless than 2.5 meters. Increasing the bandwidth to 1 GHz onlyslightly improves the performance (since the transmit poweris the same and the OFDM symbols are shorter). Now K = 5can be evaluates, since the RIS paths can be resolved. Wenotice a significant performance improvement, similar to thereflecting surface, but with larger coverage.

4) Direction of information: Finally, we visualize the direc-tions of information in the FIM for the three cases in Fig. 8.For the scatter point and the reflecting surface, these directionsare deterministic, while for the RIS, they can be optimized bychoice of the selection vector a.

VI. CONCLUSIONS

Beyond single-BS multipath-aided positioning capabilities,RISs provide a unique opportunity to control the propagationchannel and hence, to guarantee a theoretical positioningaccuracy level regardless of the occupied UE location (i.e.,

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

0

0.2

0.4

0.6

0.8

1

Figure 6. CDF of the PEB within the deployment region for K 2 {1, 2}.W = 100 MHz.

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

0

0.2

0.4

0.6

0.8

1

Figure 7. CDF of the PEB within the deployment region for K 2 {1, 2}.W = 1 GHz.

0 5 10

-5

0

5

10

15scatter point

0 5 10

-5

0

5

10

15reflector

0 5 10

-5

0

5

10

15RIS

Figure 8. Directions of information in the FIM for W = 100 MHz.The arrows show (in blue) the direction x/kxk from the BS and (in red),when the secondary path is present, the directions (x� s)/kx� sk (leftfigure), (x� xVA)/kx� xVAk (middle figure) and (x � xk)/kx� xkk(right figure).

achieving near-homogeneous localization quality-of-serviceover space in a given scene). In this paper, we have analyzedthe RIS-aided positioning problem from a FIM perspective,before introducing a simple solution to activate and optimizethe best RISs accordingly. Significant theoretical performancegains have thus been illustrated when activating one single RISalready in terms of both coverage (resp. PEB) when comparedto a single passive reflector (resp. a single passive scatterer).Obviously, the RIS-based approach is naturally hampered bylower SNR, unless the number of elements in the RIS islarge. Finally, we have also adversely observed limitations atlarge numbers of activated RISs, due to unresolved multipathcomponents.

Although far-field approximations have been made through-out the paper, the far-field distance exceeds the size of thedeployment area in the considered region. Hence future workshould explore near-field phenomena and more specifically,the possibility to exploit the curvature of wavefronts impingingonto (resp. departing from) array-based RIS (e.g., with uniform

“Triplet-Based Wireless Channel Charting,” Paul Ferrand, Alexis Decurninge, Luis G. Ordoñez, Maxime Guillaud, Globecom 2020.


Communication and sensing convergence

5G as radar• Process backscattered signal to estimate range, velocity, and angle• Needs full-duplex operation L but automatic synchronization J• OFDM waveform vs tailored waveforms (FMCW, PMCW)• Fundamental parameters:

– Bandwidth for range– Coherent integration time for Doppler– Array aperture for angle

How 6G is an enhanced radar system• Bandwidth on par with automotive radar, so can replace FMCW• Full-duplex OFDM vs new waveforms (e.g., OTFS)• Power consumption and guaranteed radar performance issues• Main benefit: ability to share information between radars• Joint radar and communication (JRC): spectrum sharing


Question: Is it radar or is it 6G ? Answer: it is both!

Barneto, Carlos Baquero, Taneli Riihonen, Matias Turunen, Lauri Anttila, Marko Fleischer, Kari Stadius, Jussi Ryynänen, and Mikko Valkama. "Full-duplex OFDM radar with LTE and 5G NR waveforms: Challenges, solutions, and measurements." IEEE Transactions on Microwave Theory and Techniques 67, no. 10 (2019): 4042-4054.

CFAR-based range-Doppler processing, the radar system is possible to track trajectories on 3-D range-velocity-angle map. The additional AOA information helps us improve target classification. The basic processing procedure includes (1) parallel reception in two receiving channels; (2) sequentially collecting number of sweeps to make a frame; (3) making range-Doppler compression by IFFT and FFT; (4) CFAR threshold for suppressing clutter and noise; and (5) combining range-Doppler data from the two channels to make 3-D range-Doppler-angle map through beamforming.

Having the 3-D range-velocity-angle map, we can track the range, velocity, and azimuth angle of detected targets from

frame to frame. By applying feature extraction algorithms, such as mean spectrogram, maximal micro-Doppler shift, and micro-Doppler signatures [5], a set of features can be collected and, then, send to a classifier, such as support vector machine (SVM), neural networks, or deep learning, to perform target classification [4].

In this paper, we will mainly describe how to use our experimental Doppler radar and based on the phase-interferometry principle to detect and track range, velocity, and angle of arrival of mini-drones

Fig. 2 – Radar system block diagram.

II. RADAR SYSTEM DESCRIPTION The C-band radar system, centered at 5.8 GHz with dual

receivers is used for the drone detection and tracking. The system consists of an RF front-end and a processing board. The heart of the processing board is the Cyclone-IV FPGA. A system block diagram is shown in Figure 2.

The system consists of an RF front-end and a Processor Board. The heart of the processor board is the Cyclone-IV FPGA. A phase-locked loop (PLL) unit in the RF front-end is controlled by the FPGA to compensate the nonlinear nature of the VCO and provide better linearity in linear FMCW with accurate sweep bandwidth. Received signals are down-converted to base-band I and Q signals and sampled by four

ADCs in the processing board. FPGA streams received I and Q samples to a PC via USB 2.0 port for further signal processing.

A patch antenna is used in the radar system via SMA RF cables. The radar transmitting power is 19 dBm and the gain for both receiving antenna is 15 dBi. For covering ±35º azimuth angle, The spacing between two receiving antennas is 0.8 wavelength.

A graphical user interface (GUI) was developed for easy-control of the waveform (FMCW/CW) and radar parameters, such as sweep bandwidth, sweep time, and sampling rate. It continuously receives data streamed from the USB port and for quick signal examination, it also provides time domain plot, range profile, range-Doppler map, etc. Data can be recorded for further processing.

Fig. 3 – Range-Doppler processing.

��,((( ��

© Drone Detection and Tracking Based on Phase-Interferometric Doppler Radar, 2018

© electronicdesign.com

Mishra, Kumar Vijay, MR Bhavani Shankar, Visa Koivunen, Bjorn Ottersten, and SergiyA. Vorobyov. "Toward millimeter-wave joint radar communications: A signal processing perspective." IEEE Signal Processing Magazine 36, no. 5 (2019): 100-114.

Liu, Fan, and Christos Masouros. "A Tutorial on Joint Radar and Communication Transmission for Vehicular Networks-Part I: Background and Fundamentals." IEEE Communications Letters (2020).


7

Location-aided communication

6G

5G

spatial'correlation shadowing interference spatial'reuse

propagation'delay distance

routing next'hop

proactive'allocation

p(x(t))

predictable' behavior'

Doppler velocity

angle'of'arrival' MIMO

path'loss distance fD = ||x(t)||/�||x� xs||�⌘

⌧ = ||x� xs||/c

h(�(x,xs)) minj

||xj � xd||

exp

✓� ||xi � xj ||

dc

◆ ||xi � xs|| < R

||xi � xj || > Rint

Di Taranto, Rocco, Srikar Muppirisetty, Ronald Raulefs, Dirk Slock, Tommy Svensson, and Henk Wymeersch. "Location-aware communications for 5G networks: How location information can improve scalability, latency, and robustness of 5G." IEEE Signal Processing Magazine 31, no. 6 (2014): 102-112.

© Henk Wymeersch and Tommy Svensson, 2020

Source: mmMAGIC


THz Sensing vs 6G?

• Sensing deeply integrated into 6G wireless communications networks– Uses the same infrastructure, tailor-made sensing signals part of the 6G air interface design, sensing reusing the communications signals, …– Mobile networks operators might be the provider of the sensing service

8

• Non-connected sensing/ sensing not connected via 6G, that benefits from the HW and SW developments in the 6G eco system– Benefitting from the R&D and efficient manufacturing and testing capabilities of 6G industry– 6G could drive availability of low cost and high performing THz HW enabling sensing in so far not economically viable domains

• Sensing enabled by 6G connectivity, but where the sensing part is not co-designed with the 6G air interface– Cooperative sensing that inherently needs 6G connectivity capabilities between the cooperative sensors– Sensing whose data is (ML/AI) processed in a cloud, in particular when low latency/high capacity/very reliable wireless connectivity is needed

to the cloud, e.g. tight control on the communications and computing in the (distributed) cloud is required for the sensing system

© Henk Wymeersch and Tommy Svensson, 2020



10

6G White Paper on Localization and Sensing

antennas, so that small devices can be packed with tens or hundreds of antennas, which will be beneficial for an-gle estimation. In addition, the high-rate communication links offered by 6G will be able to be leveraged to quickly and reliably share map and location information between different sensing devices. "is is beneficial for both ac-tive and passive sensing. To harness these benefits, chip technologies must be available that sufficiently support economies of scale. In addition, to support the develop-ment of new solutions and algorithms, suitable channel models that properly characterize the propagation of 6G waves over the hardware and the air are needed as well.

Future chip technologies: Having defined the broad initial range of relevant (new) spectrum for 6G to be as large as

0.3 - 3 THz, while regulatory bodies have recently start-ed to enable R&D up to 250 GHz, "ere is a clear need for further development of technology which will able to support the said frequency bands in a cost effective manner. One key aspect is the integration of the required technology. Currently, radio systems operating in the range of multiple 100 GHz typically include antennas and signal processing equipment, for example, which is unreasonably large to integrate into typical user equip-ment (UE). As we start to migrate towards 5G systems in the world, we see that silicon products for UE purposes have been in development for a few years while the net-work roll-out has started in a limited geographical scope only during 2019 and 2020. "is is in no small part due to the complexity of the added air interface at mmWave

Quantum Technology

New frequency bandsLarge bandwiths

MEC

Compact form factor

Higher accuracy (mm and cm)

Smartmetasurfaces

Roll, pitch, yawAI & machine learning

Dense arrays

High speed Tbps, Low latency sub-ms

Robot localizationSensing / imaging Object recognition

Low power Directional transmission

Joint comm and radar

Big data analytics and modeling of signals

AR / VR / MRRadar capabilities

Context awareness

Localization and sensing for future eHealth

Limitations in hardware

High hardware power consumption

Dark spots and blockage

Increased Interference from new services

Short range at THz

Heat problem due to very small size of THz elements & hardware

Sensor fusion

New Enablers Applications & Opportunities

Challenges

Fig. 1. Chart relating enabling technologies, 6G new application opportunities and technological challenges.

© 6G

Flagship

Communication

Localization

What is next?

Radar

Combined 6G System

Adaptive environments

AI

https://arxiv.org/abs/2006.01779

Sensing

More bandwidth,antennas

https://arxiv.org/abs/2006.01779

6G sensing and communication convergence · 2021. 4. 9. · Petteri Kela, Kari Leppanen, and Mikko Valkama. "High-efficiency device positioning and location-aware communications in

Documents