1 A Beam-Space Active Sensing Scheme for Integrated Communication and Sensing

2025-04-28 1 0 698.89KB 17 页 10玖币

侵权投诉

A Beam-Space Active Sensing Scheme for

Integrated Communication and Sensing

Applications

Saeid K. Dehkordi1, Giuseppe Caire1

Abstract

In this paper, we develop an active sensing strategy for a millimeter wave (mmWave) band In-

tegrated Sensing and Communication (ISAC) system adopting a realistic hybrid digital-analog (HDA)

architecture. To maintain a desired SNR level, initial beam acquisition (BA) must be established prior

to data transmission. In the considered setup, a Base Station (BS) Tx transmits data via a digitally

modulated waveform and a co-located radar receiver simultaneously performs radar estimation from the

backscattered signal. In this BA scheme a single common data stream is broadcast over a wide angular

sector such that the radar receiver can detect the presence of not yet acquired users and perform coarse

parameter estimation (angle of arrival, time of ﬂight, and Doppler). As a result of the HDA architecture,

we consider the design of multi-block adaptive RF-domain “reduction matrices” (from antennas to RF

chains) at the radar receiver, to achieve a compromise between the exploration capability in the angular

domain and the directivity of the beamforming patterns. Our numerical results demonstrate that the

proposed approach is able to reliably detect multiple targets while signiﬁcantly reducing the initial

acquisition time.

Index Terms

integrated sensing and communication, otfs, hybrid digital-analog beamforming, active sensing.

I. INTRODUCTION

ISAC applications have emerged as key enablers for 5G and beyond wireless systems to

deal with challenging requirements in terms of spectral efﬁciency, localization, and power con-

sumption among others [1]. In mmWave communications, it is crucial to compensate the large

1Communications and Information Theory Chair, Technical University of Berlin, Germany.

Corresponding author: s.khalilidehkordi@tu-berlin.de

arXiv:2210.04312v1 [eess.SP] 9 Oct 2022

Fig. 1: Discovery mode, where a Tx (a base station or a car) broadcasts a common message exploring a wide

angular sector

isotropic path-loss with highly directional beamforming (BF) gain. This requires fast and accurate

initial BA to be established before data transmission (see e.g. [2] and references therein). In this

work we focus on automotive applications where a transmitter (Tx) unit, e.g. a BS as a road-

side infrastructure, communicates with other vehicles. In such applications, BA for new users

entering the Field of View (FoV) is particularly challenging. Furthermore, BA is a prerequisite

for beam tracking and reﬁnement [3] of already acquired users such that the BS can continually

update the best beam for the users. In our previous works [4]–[6], we studied the joint target

detection and parameter estimation problem with a BS enhanced by a co-located Radar receiver,

using orthogonal time frequency space (OTFS), i.e. a multi-carrier modulation proposed in [7]

and applied to different Multiple-input multiple-output (MIMO) conﬁgurations (see, e.g., [8],

[9]). As an extension to the aforementioned, here we consider improved initial target detection

schemes. In this scheme, intended in the so called Discovery mode presented in [6], an OTFS

modulated signal is broadcast over a wide angular sector (ﬁg. 1). The goal of the radar receiver

is to detect the presence of targets (vehicles) that are not yet acquired, as well as estimating

their relevant parameters (angle of arrival, range, and speed).

The initial BA for HDA architectures has previously been studied in the context of a commu-

nication systems whereby the goal is to align a mobile User Equipment (UE) with a BS. A major

difference between the radar use case and the communication based case is the unavailability

of direct feedback between the two entities, i.e., the radar receiver only relies on backscattered

signals. The work of [10] builds upon that of [11], where hierarchical beamforming codebooks

are used to narrow down the angular location of the UE. The major issue for the bisection scheme

in both works is that the sensing is initialized by sensing the wide FoV with two (almost) constant

gain beam patterns, which divide the FoV in to equal sections. With the assumption of a constant

available transmit power, this in effect leads to a very low BF gain in the initial sensing stage,

especially problematic for radar uses where signal power attenuates heavily with distance,i.e.,

∝1/r4. While [10] improves the bisection scheme in [11] by applying a Posterior Matching

scheme for the selection of codewords at each level, the assumption placed on known signal-to-

noise ratio (SNR) and channel coefﬁcients is very impractical in the radar scenario. Additionally,

the same work assumes a grided approach where more reﬁned estimates, i.e., increased resolution

leads to increasingly more levels of beampatterns thus increasing the acquisition induced latency.

Perhaps the most signiﬁcant drawback of these schemes for the radar use case, is the inability

to simultaneously detect multiple users(targets). Furthermore, the work of [10] has not extended

the scheme for multiple RF chains in the HDA architecture. The presented scheme in this work

closely follows that of resource management methods in cognitive radar, especially in the spatial

domain, where the sensing output of the radar is used to improve the sensing pattern of the radar

at the consequent probing periods [12]. The main contributions of this work are summarized as

follows.

1) We propose the use of multiple randomized reduction matrices across the processing interval,

which are generated sequentially such that the matrix at block bdepends on the received

signal in previous blocks. The proposed sensing strategy, whose role is to actively sample the

FoV in the angular domain, leads to an improvement in detection probability. Additionally,

this results in an improvement in parameter estimation performance by increasing the gain

of the received signal across multiple blocks where targets are present.

2) We propose a single level codebook as compared to the hierarchical codebooks of [11].

Besides reducing the system complexity, this design ensures a constant level of BF gain at

each individual sensing level and improves the total gain via integration. Additionally, due

to the ML estimation framework, we place no limitations on the resolution of the estimates,

since as opposed to [10], the angle of arrival (AoA) estimated is not determined by the

codebook reﬁnement level.

3) The proposed scheme is easily adjustable with the number of available RF chains in a given

HDA system.

II. SYSTEM MODEL

A. Physical Model

We consider a system operating over a channel bandwidth Wat the carrier frequency fc. A

BS Tx is equipped with an uniform linear array (ULA) of Naelements with Nrf Tx RF chains

(NaNrf ), and a radar receiver co-located with the BS. For simplicity of exposition, we assume

that the Tx array and the Rx radar array coincide and that the Tx and Rx signals are separated

by means of full-duplex processing.1We consider a point target model, such that each target

can be represented by a line-of-sight (LoS) path only [14]–[16]. This model can be justiﬁed for

mmWave channels as they incur large isotropic attenuation such that all multipath components

between the BS and each target receiver disappear below the noise ﬂoor after reﬂection. By

letting φ∈[−π

2,π

2]be the steering angle and considering a ULA with λ/2spacing, the Tx/Rx

array response are given by:

[a(φ)]i=ejπ(i−1) sin(φ), i ∈1, . . . , Na(1)

Since this paper focuses on the radar processing, we consider the channel model for the backscat-

tered signal. The channel for the backscattered signal with Ptargets is given by the superposition

of Prank-1 channel matrices, each of which corresponds to the LoS propagation from the Tx

array to each target and back to the radar Rx array along the same LoS path. This results in the

Na×Natime-varying MIMO channel given by [17]

H(t, τ) =

P−1

p=0

hpa(φp)aH(φp)δ(τ−τp)ej2πνpt,(2)

where hpis a complex channel gain including the LoS pathloss and the radar cross-section

coefﬁcient [18]:

|hp|2=λ2σrcs,p

(4π)3d4

,(3)

where λ=c

fcis the wavelength, cis the speed of light, νpis the round-trip Doppler shift, τpis

the round-trip delay (time of ﬂight for a distance of dp), φpdenotes the AoA, and σrcs,p is the

radar cross section (RCS) in m2, corresponding to the p-th target. We assume that the channel

parameters {hp, φp, νp, τp}P

p=1 remain constant over the coherence processing interval TCP I of B

1Full-duplex operations can be achieved with sufﬁcient isolation between the transmitter and the (radar) detector and possibly

interference analog pre-cancellation in order to prevent the (radar) detector saturation [13].

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

1ABeam-SpaceActiveSensingSchemeforIntegratedCommunicationandSensingApplicationsSaeidK.Dehkordi1,GiuseppeCaire1AbstractInthispaper,wedevelopanactivesensingstrategyforamillimeterwave(mmWave)bandIn-tegratedSensingandCommunication(ISAC)systemadoptingarealistichybriddigital-analog(HDA)architecture.Tomain...

展开>> 收起<<

1 A Beam-Space Active Sensing Scheme for Integrated Communication and Sensing.pdf

共17页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

1 A Beam-Space Active Sensing Scheme for Integrated Communication and Sensing

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: