Opportunistic__Design_and_Experimental_Validation.pdf资源-CSDN文库

需积分: 5 26 浏览量 2024-09-08 18:01:34 上传评论收藏 2.55MB PDF 举报

资源推荐

资源详情

资源评论

IEEE INTERNET OF THINGS JOURNAL, VOL. 11, NO. 18, 15 SEPTEMBER 2024 30131

Opportunistic Passive Beamforming for

RIS-Assisted WiFi Network: System Design and

Experimental Validation

Jinsong Chen ,WeiWang , Senior Member, IEEE, and Jian Wang, Member, IEEE

Abstract—Reﬂecting intelligent surface (RIS) has been widely

used to enhance radio signals in wireless communications.

However, RIS’s capabilities, other than signal enhancement, are

rarely investigated, which restricts RIS’s role in assisting wireless

communications. In this article, we propose to apply RIS for joint

signal enhancement and collision alleviation to the contention-

based wireless access standards, e.g., IEEE 802.11 wireless

local area network (WLAN). Inspired by the capture effect’s

capability in collision alleviation, we design an opportunistic

passive beamforming (OPBF) scheme that artiﬁcially introduces

channel ﬂuctuations to manage the capture effect. We realize the

OPBF scheme through three steps, i.e., reﬂection pattern design,

random index generation, and index-guided pattern switching.

In addition, we develop a hardware prototype for validations in

WiFi networks. The experiment results show that our scheme

can effectively reduce collision rate and improve the system

throughput.

Index Terms—Capture effect, collision alleviation, IEEE

802.11, opportunistic passive beamforming (OPBF), reﬂecting

intelligent surface (RIS).

I. INTRODUCTION

ECONFIGURABLE intelligent surface (RIS), a.k.a,

programmable metasurfaces, and intelligent reﬂecting

surfaces (IRSs), is an emerging electromagnetic technology

that makes it possible for human beings to customize wireless

propagation environment [1], [2], [3], [4]. The application

of RIS to wireless communications is expected to migrate

the paradigm of “adapting to environment” to “reconﬁguring

environment” [3]. RIS comprises an array of cost-effective

and energy-effective reﬂective elements, each of which can

independently adjust the reﬂection coefﬁcient to manipulate

the reﬂected waves. Existing works have dedicated lots of

efforts to exploit RIS in improving energy efﬁciency [5], [6],

and extending signal coverage [7], [8], [9], [10]. The essence

of these applications is to build an additional link through

RIS to enhance the target user’s signal-to-noise ratio (SNR).

Manuscript received 29 February 2024; revised 9 May 2024; accepted

29 May 2024. Date of publication 7 June 2024; date of current version

6 September 2024. This work was supported by the National Natural

Science Foundation of China under Grant 62201309. (Corresponding authors:

Wei Wang; Jian Wang.)

Jinsong Chen and Jian Wang are with the School of Electronic Science

and Engineering, Nanjing University, Nanjing 210023, China (e-mail:

jinsongchen@smail.nju.edu.cn; wangjnju@nju.edu.cn).

Wei Wang is with Peng Cheng Laboratory, Shenzhen 518066, China

(e-mail: wei_wang@ieee.org).

Digital Object Identiﬁer 10.1109/JIOT.2024.3411119

However, most of the theoretical and experimental studies

focus on a tiny-toy example of RIS-assisted wireless com-

munications, where a base station (BS)/access point (AP) is

communicating with users assisted by RISs [11], [12], [13].

To conﬁgure the RIS, the interactions of BS/AP, RIS, and

users have to be introduced for RIS channel sounding and

RIS passive beamforming. The coupling structure means that

a radical upgrade of the communication protocol is needed,

which is evidently a major hindrance to RIS’s application to

the existing commercial mobile systems.

The contention-based protocol allows users to utilize the

same channel without complex precoordination. Owing to its

low complexity and cost, the most well-known contention-

based protocol, i.e., IEEE 802.11, has become a mainstream

access technology in wireless local area network (WLAN).

However, the low complexity and cost come at the expense

of performance degradation caused by collisions [14], [15],

especially in dense cases. Due to the decentralized nature of

contention-based protocols, users are self-scheduled to decide

when to initiate transmissions. In IEEE 802.11, the distributed

coordination function (DCF) is employed as the fundamental

scheme of medium access. The “listen before talk” operating

procedure is the main strategy of DCF to avoid collisions,

which, albeit effective in general, still causes heavy collisions

in high-density scenarios. Therefore, collision remains a main

performance bottleneck of contention-based networks.

In the classic 802.11 throughput analysis [14], it is assumed

that all simultaneous transmissions involved in a collision end

up with failure. Speciﬁcally, all packets are corrupted and

discharged, and stations would retransmit collided packets by

the binary exponential backoff (BEB) mechanism. However,

in practical cases, the packet with the strongest power may be

successfully detected by the receiver if its power is sufﬁciently

stronger than other packets in a collision. The phenomenon

that the receiver could correctly receive a packet from simul-

taneous transmissions is referred to as the capture effect. The

impacts of capture effect on IEEE 802.11 systems have been

widely studied in [16], [17], [18], [19], [20], [21], and [22].

These works collectively reveal the beneﬁts of the capture

effect in reducing collision rate and improving the overall

throughput of the IEEE 802.11 systems. However, some disad-

vantages come along with the performance improvement due

to the uncontrollability of the wireless channel. For example,

combined with the IEEE 802.11 DCF, the capture effect

introduces unbalanced throughput among users [17], [23],

2327-4662

 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.

See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Hong Kong Baptist University. Downloaded on September 07,2024 at 13:22:43 UTC from IEEE Xplore. Restrictions apply.

30132 IEEE INTERNET OF THINGS JOURNAL, VOL. 11, NO. 18, 15 SEPTEMBER 2024

i.e., users with better channel conditions achieve higher

throughput, which leads to fairness issues. To combat the

fairness issues, a number of interesting works manage to

artiﬁcially manipulate the capture effect on the transmitter

side [19], [24].In[19], a power control scheme is applied to

introduce the received power imbalance of users to exploit the

capture effect. Sahraei and Ashtiani [24] aimed to maximize

the saturation throughput of IEEE 802.11 wireless networks

through power randomization. However, the aforementioned

methods, developed on the transmitter side, require a radical

upgrade of the IEEE 802.11 protocol, and the incompatibility

with the current WiFi standards has prevented their usage in

practice.

In this article, we propose the opportunistic passive

beamforming (OPBF) scheme as a new paradigm to assist

contention-based wireless networks. Different from the

conventional methods, OPBF provides both signal enhance-

ment and collision alleviation functions. Inspired by the

opportunistic beamforming in [25], which utilizes dumb anten-

nas to introduce fast channel ﬂuctuations, we design the

OPBF scheme that artiﬁcially introduces channel ﬂuctuations

via RIS to manage the capture effect. The compatibil-

ity with commercial wireless communication systems, like

[26], [27], and [28], is a prior objective of this work. Shifting

from the transmitter side to the channel side, OPBF provides

a decoupled framework for commercial WiFi networks, and

can be readily applied to current infrastructures without

any upgrade. In our design, we realize the OPBF scheme

through three steps, i.e., reﬂection pattern design, random

index generation, and index-guided pattern switching. First, to

fully exploit the collision alleviation capability of the capture

effect, we develop a reﬂection pattern design scheme, in which

multiple single-beam reﬂection patterns are designed accord-

ing to the desired beamforming directions. These patterns are

assigned different activation probabilities according to factors,

such as user density, trafﬁc loads, and Quality of Service (QoS)

requirements. Then, a random index sequence is generated by

a pseudo-number generator based on activation probabilities

to guide the selection of beams. Finally, OPBF employs an

index-guided reﬂection pattern switching at each time slot. To

validate the effectiveness of our design in contention-based

wireless networks, we implement OPBF on a 12 × 12 RIS,

and evaluate its performance in a two-user IEEE 802.11a

network. Speciﬁcally, we build an IEEE 802.11 network based

on national instruments (NIs) software deﬁned radio (SDR)

devices running NI 802.11 Application Framework [29].Our

experiments show that OPBF can effectively enhance signal

strength and alleviate collision rate. We also show that OPBF

has the potential in trafﬁc steering. In addition, to address

the problem of the limited user equipment number in the

experimental study, we conduct numerical simulations for

dense networks.

Finally, the main contributions of this work are summarized

as follows.

1) OPBF can provide not only signal enhancement

but also collision alleviation. To the best of our

knowledge, OPBF is the ﬁrst RIS-enabled scheme

that artiﬁcially introduces channel ﬂuctuations to

alleviate collision in contention-based wireless

networks.

2) OPBF is ﬂexible and effective. Through managing the

beam activation probabilities, OPBF achieves ﬂexible

and effective trafﬁc steering among users. With the traf-

ﬁc steering capability, OPBF can provide differentiated

services for different users.

3) OPBF is easy to implement. OPBF decouples the

RIS and wireless networks, and can be readily

applied to commercial WiFi networks without protocol

modiﬁcations.

II. S

IGNAL ENHANCEMENT THROUGH RIS-ASSISTED

PASSIVE BEAMFORMING

In this section, we introduce the channel model of the

RIS-assisted WiFi network and study the capability of signal

enhancement through RIS-assisted passive beamforming.

A. Channel Model

We consider the infrastructure mode of WiFi, where the AP

and the users are equipped with a single antenna. A RIS with

N = N

× N

reﬂective elements is deployed to guarantee

the reliable data link between the AP and users. The channel

response between AP and user k is given by

= h

h

(1)

where h

and h

∈ C

N×1

denote the channel between the

AP and the RIS and between the RIS and user k, respectively.

The diagonal matrix  = diag{g} is the RIS response, where

g = [e

, e

,...,e

]

is the reﬂection coefﬁcients vector

that determines the reﬂection pattern. Here, w

∈ [0, 2π]is

the phase shift applied on element n, and we assume the unity

reﬂection amplitude of each element.

In addition, we assume that h

and h

follow the Rician

distribution. Thus, the channels can be represented as







+ 1



+ 1



(2)







+ 1



+ 1



(3)

where σ

= σ

)

−α

, i ∈{AR, Rk} denotes the large-

scale path-loss of the channel, and σ

denotes the reference

path-loss at d

= 1m,d

and α

represent the distance and

path-loss exponent, K

and K

are the K-factors of the

Rician channel.

and

denote the fast-fading NLoS

components of the channel, where each entry is modelled as

Gaussian random variables with distribution CN (0, 1).

and

denote the deterministic LoS component of channels,

which are position-dependent and can be further expressed by

steering vectors as

= a

(

,ϕ

)

(4)

= a

(

,ϕ

)

(5)

where a(θ

,ϕ

), a(θ

Rk,

,ϕ

) ∈ C

N×1

are the receive

response vector and the transmit response vector of RIS, θ

Authorized licensed use limited to: Hong Kong Baptist University. Downloaded on September 07,2024 at 13:22:43 UTC from IEEE Xplore. Restrictions apply.

CHEN et al.: OPPORTUNISTIC PASSIVE BEAMFORMING FOR RIS-ASSISTED WiFi NETWORK 30133

, θ

and ϕ

denote the elevation and azimuth Angles of

Arrival (AoA) and Angles of Departure (AoD).

For a uniform planar array (UPA), the response vectors

a(θ, ϕ) could be decomposed along the x-axis and y-axis,

e.g., a(θ, ϕ) = u(φ, N

) ⊗ u(ψ, N

), where ⊗ denotes

the Kronecker product, φ



= (2π/λ)d

sin θ cos ϕ, ψ



(2π/λ)d

sin θ sin ϕ represent the spatial angles, and λ is the

wavelength, with d

and d

denoting the distance between RIS

elements in x-axis and y-axis, respectively. u(φ, N) is referred

to as the steering vector function, expressed as [30]

(

φ,N

)



1, e

−jφ

, e

−j2φ

,...,e

−j(N−1)φ



. (6)

B. Single-User Signal Enhancement

With the passive beamforming vector g, a virtual LoS link

can be built with the channel response as follows [31]:

vLos

= β

(

,ϕ

)

a

(

,ϕ

)

= β

(

,ϕ

)

diag{g}a

(

,ϕ

)

= β

g (7)

where β



([σ

]/(K

+ 1)(K

+ 1))

denotes the channel gain, and c

is deﬁned as



= a

(

,ϕ

)

 a

(

,ϕ

)

= u

(

− φ

, N

)

⊗ u



− ψ

, N



(8)

and  denotes the Hadamard product. Based on (7),itis

easily derived that the optimal reﬂection coefﬁcients vector

that maximizes the received signal power of the virtual LoS

link is

= c

= u

(

− φ

, N

)

⊗ u



− ψ

, N



(9)

which is characterized by the angle tuple (θ

,ϕ

,θ

,ϕ

Speciﬁcally, the phase shift on the (n

, n

)-th element

caused by the incident signal from direction (θ

,ϕ

),is

(

)

=−

2π

sin θ

cos ϕ

+ n

sin θ

sin ϕ

). (10)

Similarly, the phase distribution to beamform the signal to the

direction (θ

,ϕ

) is

(

)

=−

2π



sin θ

cos ϕ

+ n

sin θ

sin ϕ



. (11)

To perfectly form a beam toward the desired direction, each

RIS element should compensate for the phase difference, i.e.,

(

)

= w

(

)

− w

(

)

. (12)

The RIS provides a 2-bit phase quantization, e.g., four phase

shifts: γ

∈ [0, 2π], i ∈ [1, 4]. Thus, the optimal continuous

phase w

)

is quantized to its nearest discrete phase

∗

(

)

= Q



)

2−bit



= arg min



(

)

− γ



. (13)

The quantization error w

err

)

= w

)

− w

∗

)

results in

the imperfect beam pattern as shown in Fig. 1. Although a

(a) (b)

Fig. 1. Beam patterns with continuous phases and discrete phases. (a) −45

◦

(b) 30

◦

(a) (b)

RIS

Users 1 ...K

Multiple beams

RIS

Users 1 ...K

Handover

Beam1

Beam2

Fig. 2. Passive beamforming schemes in the RIS-assisted WiFi network.

(a) JPBF. (b) OPBF.

deviation is observed from the continuous phase, the direction

of the beam has not been altered. Hence, within the context

of the 2-bit phase quantization, this particular design remains

viable.

C. Joint Multiuser Signal Enhancement

With the angular information, e.g., AoA/AoD, of a speciﬁc

user, the data link from the AP to that user can be strengthened

by directional passive beamforming using (9). However, in the

multiuser scenario, the passive beamforming design remains

an open question.

A predominant method is the joint passive beamforming

(JPBF) [32], [33] as shown in Fig. 2(a), in which the RIS

serves multiple users simultaneously. JPBF adopts a multi-

beam reﬂection pattern with each beam pointing at a certain

user. The design of a multibeam reﬂection pattern can be

formulated as the following optimization problem:

max



k=1



vLos



(14a)

s.t. w

∈{γ

,γ

}, n ∈ [1, N] (14b)



vLos



−



vLos



≤ κ, k

, k

∈ [1, K] (14c)

where the constraint (14b) is the set of quantized phase values.

The constraint (14c) will result in balanced gains among

beams in multibeam patterns, and κ is the threshold of gain

imbalance. In the JPBF, RIS works as a scatterer that radiates

the energy from the AP to users or reversely as a gatherer that

collects the energy from users to the AP.

The pattern design in the JPBF is quite a challenging task.

First, the optimization with the discrete variable is a nonconvex

problem. Second, the variable space grows exponentially

with the number of RIS elements, which is computationally

prohibitive for even a small-sized RIS. Therefore, the online

Authorized licensed use limited to: Hong Kong Baptist University. Downloaded on September 07,2024 at 13:22:43 UTC from IEEE Xplore. Restrictions apply.