AdaComm Tracing Channel Dynamics for Reliable Cross -Technology Communication Weiguo Wang1 Xiaolong Zheng2 Yuan He1 Xiuzhen Guo1

2025-04-27 0 0 716.02KB 9 页 10玖币

侵权投诉

AdaComm: Tracing Channel Dynamics for Reliable

Cross-Technology Communication

Weiguo Wang1, Xiaolong Zheng2, Yuan He1, Xiuzhen Guo1

1School of Software and BNRist, Tsinghua University

2School of Computer Science, Beijing University of Posts and Telecommunications

wwg18@mails.tsinghua.edu.cn, zhengxiaolong@bupt.edu.cn,

he@greenorbs.com, guoxz16@mails.tsinghua.edu.cn

Abstract—Cross-Technology Communication (CTC) is an

emerging technology to support direct communication between

wireless devices that follow different standards. In spite of

the many different proposals from the community to enable

CTC, the performance aspect of CTC is an equally important

problem but has seldom been studied before. We find this

problem is extremely challenging, due to the following reasons:

on one hand, a link for CTC is essentially different from a

conventional wireless link. The conventional link indicators like

RSSI (received signal strength indicator) and SNR (signal to noise

ratio) cannot be used to directly characterize a CTC link. On

the other hand, the indirect indicators like PER (packet error

rate), which is adopted by many existing CTC proposals, cannot

capture the short-term link behavior. As a result, the existing

CTC proposals fail to keep reliable performance under dynamic

channel conditions. In order to address the above challenge, we

in this paper propose AdaComm, a generic framework to achieve

self-adaptive CTC in dynamic channels. Instead of reactively

adjusting the CTC sender, AdaComm adopts online learning

mechanism to adaptively adjust the decoding model at the CTC

receiver. The self-adaptive decoding model automatically learns

the effective features directly from the raw received signals that

are embedded with the current channel state. With the lossless

channel information, AdaComm further adopts the fine tuning

and full training modes to cope with the continuous and abrupt

channel dynamics. We implement AdaComm and integrate it

with two existing CTC approaches that respectively employ

CSI (channel state information) and RSSI as the information

carrier. The evaluation results demonstrate that AdaComm can

significantly reduce the SER (symbol error rate) by 72.9% and

49.2%, respectively, compared with the existing approaches.

INTRODUCTION

The ever-developing Internet of Things (IoT) brings the

widespread deployments as well as the rich diversity of

wireless technologies [1]–[3]. To directly interconnect the

heterogeneous devices that follow different wireless technolo-

gies, Cross-Technology Communication (CTC) is proposed

to enable the direct communication between incompatible

devices without extra hardware.

Despite the tremendous advances, existing CTC approaches

usually focus on enabling the communication between incom-

patible technologies. How to maintain the reliable performance

in the intrinsically dynamic channels has not received enough

attention. To convey data, existing CTC techniques usually

explore the mutually accessible information carrier such as the

energy and timing of packet transmissions [4]–[7], the state

variations of overlapped channels [8], [9], and the originally

incompatible but similar signals [10]–[12]. Since most CTC

leverages the signal patterns rather than the underlying raw

signals to convey data, the CTC links significantly differ from

the links in traditional wireless communication. Traditional

link quality indicators such as RSSI or CSI used by WiFi to

learn the channel state cannot reflect the quality of a CTC

link. So far there is not a link quality indicator to accurately

describe the CTC’s channel state.

Without an accurate CTC link indicator to learn the current

channel state, to cope with channel dynamics is a fundamental

but challenging task for CTC. Most of the existing CTC

approaches adopt indirect indicators, e.g. Packet Error Rate

(PER), to detect the changes of channel state. When there

is a significant variation in the PER, the CTC approaches

may reactively control the sender’s encoding behavior, so

as to enhance the features of encoded signals received by

the receiver. For example, WiZig [6] extends the symbol

window length and enlarges the differences of adjacent en-

coded amplitudes to enhance the features of CTC symbols.

FreeBee [4] increases the number of beacon repetitions per

symbol in the noisy channel, thus improving the highest fold

sum. ZigFi [8] controls the transmission power to maintain

Signal-to-Interference-plus-Noise-Ratio (SINR) perceived by

the receiver so that the CSI of ZigFi symbols still satisfy the

decoding model.

The reactive adjustments of CTC against channel dynamics,

unfortunately, suffer performance degradation and even failure

in practice. First, the indirect indicators of channel state only

reflect the long-term average channel quality but are insensitive

to the short-time channel dynamics. Therefore, adjusting CTC

according to those indicators can only achieve sub-optimal

performance. Second, existing decoding methods usually adopt

the threshold or machine learning model as the pattern recog-

nition methods. But features such as the RSSI threshold and

CSI variation predefined by the feature-based decoding model

cannot accurately describe the channel dynamics. For example,

the PHY information such as CSI can be affected by both the

intentional CTC transmissions and the channel-related factors

like multipath. The predetermined statistical features are not

necessarily effective to cover all possible cases, due to the

uncertainty of channel dynamics.

In order to address the above problems, in this paper we

propose AdaComm, a general and lightweight online adaptive

CTC framework that automatically adjusts the decoding model

to maintain reliable communication performance in dynamic

channel conditions. Instead of enhancing the features of signals

from the sender, we propose a self-adapting decoding model

at the receiver side, which traces the channel state to improve

the decoding reliability. We directly use raw received data

as input and avoid the information loss caused by manual

feature extraction. Our model automatically extracts the ef-

fective features to distinguish between the intended impact of

CTC modulation and the channel dynamics. We also design

an online learning mechanism that leverages the correctly

decoded CTC data to update the decoding model without extra

cost of data collection, which is called fine tuning. With fine

tuning, AdaComm is able to cope with continuous changes of

the channel state. To deal with model failures caused by abrupt

channel changes, AdaComm integrates a full training mode

that retrains the decoding model with the newly collected

training sequences. To reduce the cost of data collection for

full training, we devise a data augmentation method to obtain

sufficient training data with only limited size of the training

sequence. The main contributions of this work are summarized

as follows.

•

We propose AdaComm, a general online learning CTC

framework to maintain reliable performance in dynamic

channels. AdaComm integrates a lightweight decoding

model that takes the information of channel state into con-

sideration and automatically extracts decoding features.

•

We design fine tuning and full training modes to cope

with continuous and abrupt channel changes. In fine

tuning, we use the correctly decoded CTC data to update

the decoding model. In full training, we propose a data

augmentation method to reduce the cost of collecting

online training data.

•

We implement AdaComm on both CSI-based and RSSI-

based CTC and evaluate its performance in various en-

vironments. The experiment results show that AdaComm

can reduce the SER by 72.9% and 49.2% for CSI-based

and RSSI-based CTC, respectively.

The rest of this paper is organized as follows. We present

the related work in Section II. In Section III, we investigate

the performance of existing CTC in a dynamic environment

and analyze the causes of performance degradation. Section

IV presents the design of AdaComm. We evaluate the perfor-

mance of AdaComm in Section V and conclude our work in

Section VI.

II.

RELATED WORK

Cross-Technology Communication (CTC) has been devel-

oping rapidly and applied for channel coordination among

heterogenous technologies [13], [14]. The common idea of

CTC is building the mutually accessible information carrier

with existing hardware to convey data. One of the most

common information carriers is the energy of packet trans-

missions. ESense [5] modulates symbols by packet lengths

and accomplishes CTC from WiFi to ZigBee. HoWiEs [15]

improves Esense by using combinations of WiFi packets.

Validation Dataset

Fig. 1: Decoding accuracy in the dynamic environment.

GSense [16] embeds symbols into gaps between customized

packet preambles. B2W2 [17] mimics the DAFSK for com-

munication from BLE to WiFI. C-Morse [18] constructs the

radio energy patterns with Morse Coding. WiZig [6] improves

the throughput by using multiple amplitudes. FreeBee [4]

leverages the transmitting timing of beacon packets as the

information carrier. DCTC [19] utilizes the transmitting timing

of application packets to encode data. Another information

carrier is the channel state. ZigFi [8] uses the impacts of

ZigBee packets on Channel State Information (CSI) to convey

data. Recently, researchers utilize physical layer information

to achieve high-speed CTC. SymBee [9] creates distinguishing

phase patterns on WiFi receiver with special ZigBee pay- load.

WEBee [10] and BlueBee [11] emulate the signal of another

technology in the payload. XBee [20] utilizes bit patterns to

decode ZigBee packets at BLE. TiFi [21] utilizes

backscattered harmonic to achieve CTC between WiFi and

RFID. WIDE [12] emulates the phase shift of the receiver

directly to achieve digital emulation. Meanwhile, CTC has

many practical applications, such as channel coordination [22]

and time synchronization [23].

Despite the tremendous advances, existing CTC approaches

usually focus on enabling the communication between incom-

patible technologies. However, how to maintain the reliable

performance in the intrinsically dynamic channels has not

received enough attention. Existing methods only reactively

bear the channel dynamics and heuristically sacrifice perfor-

mance of throughput to lower the SER by retransmission [4],

[10], increasing symbol window length [6], [7], controlling

transmission power [8]. Different from existing methods, we

directly include the channel state into our decoding model and

exploit online learning to continuously adapt to the channel

dynamics.

III.

MOTIVATION

In this section, we study the performance of existing CTC

techniques in dynamic environments and further investigate

the challenges that CTC encounters in dynamic environments.

Existing CTC methods usually use the threshold or the

machine learning model as the decoding model to decode

CTC symbols. Without losing generality, we investigate the

performance of ZigFi [8] (a CTC from ZigBee to WiFi) as the

example to show the impacts of dynamic environment on CTC.

The basic idea of ZigFi is using the presence and absence of

ZigBee packets to modulate the overlapped channel to transmit

Time

12:30(0)

12:45(1)

13:00(2)

13:15(3)

13:30(4)

13:45(5)

14:00(6)

14:15(7)

14:30(8)

14:45(9)

12:30(0) 12:45(1) 13:00(2) 13:15(3) 13:30(4) 13:45(5) 14:00(6) 14:15(7) 14:30(8) 14:45(9)

0.913

0.833

0.854

0.673

0.753

0.611

0.630

0.580

0.642

0.611

0.956

0.942

0.404

0.611

0.620

0.607

0.480

0.750

0.544

0.826

0.930

0.755

0.911

0.770

0.429

0.929

0.790

0.587

0.809

0.901

0.888 0.921 0.936 0.901

0.946

0.515

0.548

0.485

0.546

0.923

0.895

0.869

0.728

0.705

0.960

0.895

0.607

0.822

0.610

0.717

0.914

0.710

0.640

0.679 0.544 0.538

0.606 0.886 0.843

0.543

0.670

0.937

0.589

0.584

0.918

0.956

0.754

0.928

0.579

0.507

0.503

0.720

0.562

0.541

0.795

0.904

0.973

0.856

0.734

0.551

0.524

0.604

0.660

0.668

0.741

0.825

0.853

0.921

0.573

0.474

0.564

0.568

0.719

0.761

0.683

0.793

0.731

0.678

0.891

Training Dataset

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：

AdaComm:TracingChannelDynamicsforReliableCross-TechnologyCommunicationWeiguoWang1,XiaolongZheng2,YuanHe1,XiuzhenGuo11SchoolofSoftwareandBNRist,TsinghuaUniversity2SchoolofComputerScience,BeijingUniversityofPostsandTelecommunicationswwg18@mails.tsinghua.edu.cn,zhengxiaolong@bupt.edu.cn,he@greenorbs.co...

展开>> 收起<<

AdaComm Tracing Channel Dynamics for Reliable Cross -Technology Communication Weiguo Wang1 Xiaolong Zheng2 Yuan He1 Xiuzhen Guo1.pdf

共9页,预览2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！

AdaComm Tracing Channel Dynamics for Reliable Cross -Technology Communication Weiguo Wang1 Xiaolong Zheng2 Yuan He1 Xiuzhen Guo1

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: