03SIGCOMM2013Session8-12_sigcomm发一篇很厉害吗资源-CSDN文库

共18个文件

pdf：18个

2013

5星 · 超过95%的资源需积分: 9 28 浏览量 2013-09-20 14:57:42 上传评论收藏 28.23MB ZIP 举报

**SIGCOMM 2013会议详解** SIGCOMM（Special Interest Group on Data Communication）是计算机通信领域的顶级学术会议，每年都会吸引全球的研究者们分享他们在网络通信、互联网架构、网络性能优化、移动通信以及新兴网络技术等方面的最新研究成果。2013年的SIGCOMM会议，即"03 SIGCOMM 2013 Session8-12"，涵盖了广泛的主题，旨在推动网络研究的发展和技术创新。在这些会议记录中，我们可以预见到当时的一些关键研究焦点和热点问题。虽然具体的论文内容未给出，但根据会议的分组和主题，我们可以推测一些可能涵盖的知识点： 1. **网络架构与设计**：SIGCOMM 2013可能涉及了网络架构的改进，包括数据中心网络的新设计、互联网协议的优化、未来网络架构的探讨等。研究人员可能会提出新的拓扑结构、路由算法或流量工程策略来提高网络效率和可扩展性。 2. **网络性能分析**：会议可能会深入讨论网络性能的度量、建模和预测。这可能包括延迟、带宽利用率、丢包率等方面的研究，以及如何通过测量和分析这些指标来改善网络服务。 3. **无线与移动通信**：随着移动设备的普及，移动通信和无线网络的研究成为热点。可能的议题包括移动网络的优化、无线频谱管理、多跳网络、自组织网络（MANETs）以及4G/LTE技术的挑战和解决方案。 4. **网络安全与隐私**：网络安全是永恒的话题，可能的论文会涵盖新的攻击手段、防御策略、隐私保护技术，以及如何在确保用户隐私的同时提供高效的服务。 5. **内容分发与云计算**：随着CDN（内容分发网络）和云计算的兴起，如何有效地存储、检索和传输大量数据成为了研究的重点。可能涉及的议题包括数据存储策略、分布式计算、云网络的性能优化等。 6. **绿色通信与能源效率**：环保意识的增强使得网络的能源效率成为一个重要的研究方向。会议可能包含了关于降低网络能耗、绿色数据中心的设计以及能源自给网络的研究。 7. **软件定义网络（SDN）与网络功能虚拟化（NFV）**：2013年是SDN和NFV开始崭露头角的时期，可能有论文探讨这些新技术如何改变网络的管理和运营方式，以及它们在提高灵活性和可编程性方面的潜力。 8. **物联网与传感器网络**：随着物联网的快速发展，传感器网络的设计、部署以及数据处理技术也是关注的焦点。可能的研究涉及物联网的安全性、数据融合以及低功耗通信技术。虽然无法详述每篇论文的具体内容，但这些主题为我们提供了一个窗口，让我们了解2013年SIGCOMM会议上讨论的核心议题和当时的科研趋势。这些研究成果对后来的网络技术发展产生了深远影响，推动了网络行业的创新和进步。

资源推荐

资源详情

资源评论

收起资源包目录

SIGCOMM 2013 Session8-12.zip （18个子文件）

Session8-12

Session 8 Network Measurement

p243.pdf 701KB

p267.pdf 742KB

p255.pdf 1.13MB

Session 11 Wireless Communication 2

p375.pdf 29.32MB

p387.pdf 2.5MB

p399.pdf 3.01MB

p363.pdf 2.98MB

Session 12 Data Center Networks 2

p435.pdf 1.35MB

p411.pdf 969KB

p447.pdf 1.72MB

p423.pdf 970KB

Session 9 Privacy

p303.pdf 561KB

p279.pdf 2.37MB

p291.pdf 913KB

p315.pdf 454KB

Session 10 Applications and Resource Allocation

p327.pdf 1.01MB

p339.pdf 1.68MB

p351.pdf 1007KB

Full Duplex Radios

Dinesh Bharadia

Stanford University

dineshb@stanford.edu

Emily McMilin

Stanford University

emcmilin@stanford.edu

Sachin Katti

Stanford University

skatti@stanford.edu

ABSTRACT

This paper presents the design and implementation of the ﬁrst in-

band full duplex WiFi radios that can simultaneously transmit and

receive on the same channel using standard WiFi 802.11ac PHYs

and in typical SNR regimes achieve close to the theoretical dou-

bling of throughput. Our design uses a single antenna for simul-

taneous TX/RX (i.e., the same resources as a standard half duplex

system). We also propose novel analog and digital cancellation tech-

niques that cancel the self interference to the receiver noise ﬂoor, and

therefore ensure that there is no degradation to the received signal.

We prototype our design by building our own analog circuit boards

and integrating them with a fully WiFi-PHY compatible software ra-

dio implementation. We show experimentally that our design works

robustly in noisy indoor environments, and provides close to the ex-

pected theoretical doubling of throughput in practice.

Categories and Subject Descriptors

C.2.1 [Computer Communication Networks]: Network Architec-

ture and Design—Wireless communication

General Terms: Algorithms, Design, Experimentation, Performance

Keywords: Full Duplex, Interference Cancellation, Non-linear Can-

cellation

1. INTRODUCTION

“It is generally not possible for radios to receive and

transmit on the same frequency band because of the in-

terference that results."

(Andrea Goldsmith, Wireless Communications [8])

The above quote captures a long-held assumption in wireless system

design that radios have to operate in half duplex mode (i.e., either

transmit or receive but not both simultaneously) on the same chan-

nel. Recent work has attempted to invalidate this assumption. Re-

searchers at Stanford [11, 3], Rice [7, 6] and several other groups in

industry and academia [14, 1] have proposed various designs to build

in-band full-duplex radios. Full duplex, if possible, has tremendous

implications for network design, not least of which is the fact that

cellular networks could cut their spectrum needs by half. For ex-

ample, LTE uses equal width separate uplink and downlink channels

to enable radios to achieve full duplex. With an in-band full-duplex

system we could use a single channel to get the same performance.

Permission to make digital or hard copies of all or part of this work for personal or

classroom use is granted without fee provided that copies are not made or distributed

for proﬁt or commercial advantage and that copies bear this notice and the full cita-

tion on the ﬁrst page. Copyrights for components of this work owned by others than

ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re-

publish, to post on servers or to redistribute to lists, requires prior speciﬁc permission

and/or a fee. Request permissions from permissions@acm.org.

August 12–16, 2013, Hong Kong, China.

Consequently, the problem has attracted signiﬁcant attention, both

from industry and academia and has spurred signiﬁcant follow-up

work.

To achieve full duplex, a radio has to completely cancel the sig-

niﬁcant self-interference that results from its own transmission to

the received signal. Since WiFi signals are transmitted at 20dBm

(100mW) average power, and the noise ﬂoor is around −90dBm, the

transmit self-interference has to be canceled by 20dBm−(−90dBm) =

110dB to reduce it to the same level as the noise ﬂoor and render it

negligible. If self-interference is not completely canceled, any resid-

ual self-interference acts as noise to the received signal and reduces

SNR and consequently throughput. For example, if the received sig-

nal’s SNR without full duplex is 25dB but is reduced to 5dB due

to 20dB residual self-interference, then the throughput with full du-

plex is that achieved using two 5dB SNR links. This is signiﬁcantly

worse than using the original half duplex link with 25dB SNR and it

is better to turn off full duplex in this case. To sum up, the amount

of self-interference cancellation dictates overall throughput and is a

ﬁgure of merit for any full-duplex design.

Prior designs have made signiﬁcant progress on the self-interference

cancellation problem [11, 5, 3]. However the best performing prior

designs can at best provide 85dB of cancellation, which still leaves

about 25dB of residual self-interference and therefore reduces the

SNR of each direction of the full duplex link by 25dB. A calculation

similar to the previous paragraph’s shows that to see throughput ben-

eﬁts with these full-duplex designs, the half-duplex SNR of the link

has to be extremely high (45dB or higher). In terms of range, the

two nodes would have to be closer than 5m to see such high SNRs.

Outside this range, it is better to turn off full duplex and use the tra-

ditional half duplex mode. To be fair however, these designs were

intended for low-power, narrow-band, ﬁxed rate protocols such as

Zigbee where 85dB of self-interference cancellation is sufﬁcient for

full duplex. WiFi is far more demanding both in terms of bandwidth

as well as cancellation.

Prior designs also need to have at least two antennas [11, 5] in

place of the one used by half duplex systems (one each for transmit

and receive and possibly more [3]). However, with two or more an-

tennas, the argument for full duplex becomes weaker since the same

doubling of capacity could be obtained by using the two antennas as

MIMO antennas to spatially multiplex two independent packets in

half duplex mode instead of using them for full duplex.

In this paper, we present the design and implementation of a full

duplex WiFi radio that uses a single antenna

and delivers close to

Picasso [10] uses a single antenna, but it only allows the radio to

simultaneously transmit and receive on different adjacent channels.

Hence it fails to address the much harder problem of simultaneous

TX/RX on the same channel. Our system does address this chal-

lenge, and offers novel and higher performance analog and digital

cancellation techniques compared to Picasso.

375

SIGCOMM’13,

the theoretical doubling of throughput under all typical SNR regimes.

Our key technical contributions are novel self-interference cancella-

tion circuits and algorithms that provide the required 110dB of self

interference cancellation for standard WiFi signals and thus elim-

inate all self interference to the noise ﬂoor. Our design is wide-

band: it works with the highest bandwidths (80MHz) and data rates

used by the latest 802.11ac PHY in the 2.4GHz spectrum. We also

experimentally demonstrate a complete full-duplex communication

link which uses the full WiFi PHY (OFDM, constellations up to

256QAM and all the channel coding rates) and achieves close to the

theoretically expected doubling of throughput. To the best of our

knowledge, this is the ﬁrst working implementation of a complete

WiFi PHY single-antenna full-duplex link.

The reader might be wondering why full duplex is hard to real-

ize. After all, as the sender knows the signal being transmitted, sub-

tracting it should be relatively simple to implement. One of the key

insight in this work is that in fact the radio does not know what it

is transmitting. What it does know is the clean digital representa-

tion of the signal in baseband. However, once the signal is converted

to analog and up-converted to the right carrier frequency and trans-

mitted, the transmitted signal looks quite different from its baseband

incarnation. The numerous analog components in the radio TX chain

distort the signal in both linear and non-linear ways (analog circuits

will create cubic and higher order components of the signal for ex-

ample), add their own noise (e.g., power ampliﬁers add transmitter

noise), are slightly inaccurate (e.g., your oscillator is tuned slightly

off 2.45GHz), or delay it by different amounts at different frequen-

cies and so on. In effect the transmitted signal is a complicated non-

linear function of the ideal transmitted signal along with unknown

noise. Unsurprisingly, naively subtracting a “known” baseband ver-

sion of the transmit signal without accounting for all these analog

distortions does not work. As we will show in Sec 5 prior designs

fail to account for these distortions and hence are limited to at best

85dB of cancellation.

This paper makes two key contributions over all prior work in this

space. First, we design dynamic algorithms to estimate the distor-

tions introduced by analog circuits and accurately model the actual

self-interference being experienced by the received signal. Second,

we design a novel programmable analog cancellation circuit using

off-the-shelf components that allows us to implement the above al-

gorithm in “analog” and dynamically cancel the self-interference.

Such analog cancellation prevents receiver saturation from strong

self-interference and allows us to use commodity radios. However,

the analog cancellation stage does not completely cancel the self-

interference. We complement it with a novel digital cancellation al-

gorithm and implementation that cancels any remaining self-interference.

Our digital cancellation algorithm differs from all prior work because

it not only models the linear distortions, but also non-linear effects

and other special effects such as oscillator noise. Thus, overall we

use a hybrid analog-digital design that successfully models all linear

and non-linear distortions as well as transmitter noise.

We implement our design via a combination of circuit designs and

software implementations. Our analog cancellation is implemented

on a PCB that we designed and populated using off-the-shelf compo-

nents. We integrate our board with an off-the-shelf antenna and soft-

ware radio transceiver [16, 15] based on test equipment from Rohde-

Schwarz (RS) as well as on commodity WARP radios. We also im-

plement our digital cancellation algorithms as well as a fully WiFi

compliant PHY layer based on OFDM, supporting constellations up

to the standard required 256QAM and all the channel coding rates.

We deployed and evaluated our system in an indoor and noisy ofﬁce

environment in the 2.4GHz ISM band, operating the WiFi PHY over

Figure 1: What we think we are transmitting in digital on the left side, and

what the radio actually transmitted on the right side. The actual transmitted

signal differs signiﬁcantly from the two tones generated in digital baseband.

Note transmitter noise and harmonics are generated in addition to the two

main transmitter tones.

the 80MHz bandwidth on RS radios, and over the 20MHz bandwidth

using WARP radios.

Our experiments demonstrate that our design delivers on the promise

of full duplex. Under typical indoor deployment scenarios, our sys-

tem delivers a median throughput gain of 87% in practice with WiFi

radios which is close to the theoretically expected 2×. Looking into

the cancellation itself, we show that our design consistently deliv-

ers the required 110dB of cancellation in a dense indoor ofﬁce en-

vironment for both the RS 80MHz radios as well as the commod-

ity 20MHz WARP radios. The system is robust to environmental

changes, reﬂections, and can handle all the different constellations

used in WiFi. We compare against the best known prior full duplex

approaches [11, 7] and show experimentally that they can at best

deliver 85dB of cancellation and therefore reduce the SNR of the

received signal by at least 25dB.

2. THE PROBLEM

Full duplex, in theory, should be simple to accomplish. After all,

we know the signal we are transmitting and we are only designing

circuits and algorithms to subtract it from the received signal. The

intuition follows from the conventional abstraction that the analog

radio (also known as the RF front-end) is a black-box that takes the

digital baseband signal, converts it to analog, up-converts it to the

carrier frequency, scales it to the right power and sends it. In other

words, the assumption has been that the radio preserves the original

baseband signal except for power scaling and frequency shifting. In

practice this abstraction turns out to be incorrect. Radios in fact sig-

niﬁcantly distort the signal being transmitted, relative to the digital

baseband representation.

To demonstrate the distortions, we use the following experiment

throughout this section. We take a software radio transceiver [16, 15]

and send the following signal: two tones at 2.449GHz and 2.451GHz.

In other words, we are sending an extremely simple signal, two sine

waves with frequencies 1MHz away from the carrier frequency of

2.45GHz. We do this by creating a digital baseband signal with sam-

ples of the sine waves at −1MHz and 1MHz which the radio then

up-converts to 2.45GHz and ampliﬁes to 20dBm average transmit

power (the power used by WiFi radios). We then compare the signal

output of the antenna to what we would ideally expect if the radio did

not introduce any distortions. This experiment serves as some sort of

lower bound on the quality of radios. If radios cannot transmit even

this simplest of signals without distortion then more complex signals

such as WiFi are likely to be signiﬁcantly distorted. Fig. 1 plots the

ideal and actual transmitted signals’ spectra that resulted from our

experimental set-up (we ensured that this was a clean environment

with no other interference present in the environment at the time of

the experiment).

376

Ideally, we expect to see only two tones at 2.451GHz and 2.449GHz

as shown on the left side of Fig. 1. However in the transmitted sig-

nal, whose spectrum is plotted on the right side of Fig. 1, we can

easily see that there are several other distortions present in addition

to the two main tones that were transmitted. These distortions can be

classiﬁed into two major categories:

1. Linear Components: This corresponds to the two main tones them-

selves which are attenuated and could consist of reﬂections from the

environment. These are linear components because the received dis-

tortion can be written as a linear combination of different delayed

copies of the original two tones.

2. Non-Linear Components: These components are created because

radio circuits can take in an input signal x and create outputs that

contain non-linear cubic and higher order terms such as x

, x

These higher order signal terms have signiﬁcant frequency content

at frequencies close to the transmitted frequencies, which directly

correspond to all the other harmonics we see on the right side of

Fig. 1. Harmonics, as the name suggests, are signal distortions

which occur at equally spaced frequency intervals from the trans-

mitted frequencies. As the right side of Fig. 1 shows, we see spikes

at frequencies 2.447GHz and 2.453GHz, that are spaced 2MHz

apart from the two transmitted tones 2.451GHz and 2.449GHz, on

either side.

3. Transmitter Noise: The general increase we see in the base signal

level which we can clearly see on the sides of the two main tones is

noise from the radio transmitter. A radio will of course always have

noise, which works out to a noise power level of -90dBm [15]). But

as we can see, the power at the side-bands is signiﬁcantly higher, on

the level of −50dBm, or 40dB higher than the thermal noise ﬂoor.

This extra noise is being generated from high power components in

the radio transmitter such as power ampliﬁers. In the radio literature

this is referred to as broadband noise [12]. Further radios have phase

noise generated by local oscillator (LO), which is typically of level

of −40dBm, or 50dB above (not seen in the Fig. 1).

2.1 Requirements for Full Duplex Designs

The above analysis suggests that any in-band full duplex system

has to be able to cancel all the above distortions in addition to the

main signal component itself, since all of these are within the fre-

quency band we are transmitting and receiving on and act as strong

self-interference to the received signal itself. In this section, we dis-

cuss how strong each of these components are for typical transceivers,

and what are the requirements for full duplex. We will state all self-

interference power levels relative to the receiver noise ﬂoor. The

reason is that to implement full duplex, we need to cancel any self-

interference enough so that its power is reduced to the same level as

the receiver noise ﬂoor. There is no point in canceling beyond that

since we won’t see any beneﬁts — the received signal’s SNR will

then be dictated anyway by the receiver noise ﬂoor which cannot be

canceled or reduced, just as it is today in half duplex radios.

We use similar experiments for OFDM-wideband signals to quan-

tify the power levels of the different distortions, shown in the left

side of Fig. 2. In a typical WiFi radio using 80MHz bandwidth,

the receiver has a noise ﬂoor of −90dBm (1 picowatt). First, since

the main signal component is being transmitted at 20dBm (100mW),

self-interference from the linear main component is 20 − (−90) =

110dB above the receiver noise ﬂoor. Second, we observed exper-

imentally that the non-linear harmonics are at −10dBm, or 80dB

above the receiver noise ﬂoor. Finally, the transmitter noise is at

−40dBm, or 50dB above the receiver noise ﬂoor. Note that these

numbers are consistent with other RF measurement studies reported

in the literature [21] for standard WiFi radios.

There are four takeaways from the above analysis:

• Any full duplex system needs to provide 110dB of linear self-

interference cancellation to reduce self-interference to the receiver

noise ﬂoor. This will ensure that the strongest component (the main

signal) which is 110dB above the noise ﬂoor will be eliminated.

• A full duplex system has to reduce non-linear harmonic components

that are 80dB above the noise ﬂoor, so any full duplex technique has

to provide at least 80dB of non-linear self-interference cancella-

tion.

• Transmitter noise is by deﬁnition noise and is random. In other

words, we cannot infer it by any algorithm. Hence the only way to

cancel transmitter noise is to get a copy of it where it is generated,

i.e. in the analog domain and cancel it there. This implies any

full duplex system has to have an analog cancellation component

that provides at least 50dB of analog noise cancellation so that

transmitter noise is reduced to below the receiver noise ﬂoor.

• A ﬁnal constraint is that RX chains in radios get saturated if the

input signal is beyond a particular level that is determined by their

ADC resolution. Assuming a 12 bit ADC resolution typically found

in commodity WiFi radios, we have a theoretical 72dB of dynamic

range, which implies that the strongest signal level that can be input

to the radio relative to the receiver noise ﬂoor is −90dBm+72 =

−18dBm. However, in practice it is necessary to leave 2 bits worth

of margin, i.e a 12 bit ADC should be used as if it is a 10 bit ADC to

reduce quantization noise. So the maximum input signal level can

be −90dBm+60 = −30dBm. Since in WiFi, the transmitted self-

interference can be as high as 20dBm, a full duplex system needs

to have an analog cancellation stage that provides 60dB of self-

interference reduction (we keep a further 10dB margin for OFDM

PAPR where instantaneously an OFDM signal’s power level can

rise 10dB above the average power).

To sum up, any full duplex design needs to provide 110dB of linear

cancellation, 80dB of non-linear cancellation, and 60dB of analog

cancellation.

Transmitted Signal

60 dB of

Analog

Cancellation

Receiver

80 dB

Harmonics

50 dB

Transmitter

noise

110 dB

Main

Signal

50 dB of

Digital

Cancellation

-90

-40

-10

Power in dBm

-90 dBm Receiver Noise floor

10 dB PAPR

Receiver

Saturation

-30

Figure 2: On the left hand side we see transmitted signal with sub-

components. On the right hand side we see how this impacts the requirements

of analog and digital cancellation.

2.2 Do Prior Full Duplex Techniques Satisfy these

Requirements?

There are two state-of-the-art designs: ones which use an extra

transmit chain to generate a cancellation signal in analog [6] and

ones which tap the transmitted signal in analog for cancellation [11,

3]; both use a combination of analog and digital cancellation. Note

that all these designs use at least two antennas for transmit and re-

ceive instead of the normal single antenna, and the antenna geometry

ones use more than two.

Designs which use an extra transmitter chain report an overall to-

tal of 80dB of self-interference cancellation (we have been able to

reproduce their results experimentally). Of this, around 50dB is ob-

tained in the analog domain by antenna separation and isolation be-

tween the TX and RX antennas of around 40cm (the designs also

377

assume some form of polarization/metal shielding between the TX

and RX antennas to achieve 50dB isolation). Note that this 50dB

reduction applies to the entire signal, including linear and non-linear

components as well as transmitter noise since it is pure analog sig-

nal attenuation. Next, these designs also use an extra transmit chain

to inject an antidote signal [6, 9] that is supposed to cancel the self-

interference in analog. However, the antidote signal only models

linear self-interference components and does not model non-linear

components. Further, it is incapable of modeling noise because by

deﬁnition noise is random and cannot be modeled. Overall this extra

cancellation stage provides another 30dB of linear self-interference

cancellation in the best case. Thus, these designs provide 80dB of

linear cancellation, 50dB of non-linear cancellation and 50dB of ana-

log noise cancellation, falling short of the requirements by 30dB for

the non-linear components. Hence if full duplex is enabled over links

whose half duplex SNR is 30dB or lower, then no signal will be de-

coded. Further to see any throughput improvements with full duplex,

the half duplex link SNR would have to be greater than 50dB.

The second design [11] gets a copy of the transmitted analog signal

and uses a component called the balun (a transformer) in the analog

domain to then create a perfectly inverted copy of the signal. The

inverted signal is then connected to a circuit that adjusts the delay

and attenuation of the inverted signal to match the self interference

that is being received on the RX antenna from the TX antenna. We

show experimentally in Sec. 5, that this achieves only 25dB of ana-

log cancellation, consistent with the prior work’s results. The can-

cellation is limited because this technique is very sensitive to and

requires precise programmable delays with resolution as precise as

10picoseconds to exactly match the delay experienced by the self-

interference from the TX to the RX antenna. Such programmable

delays are extremely hard to build in practice, at best we could ﬁnd

programmable delays with resolution of 100−1000picoseconds and

these were in fact the ones used by the prior design [11]. Hence

the cancellation circuit is never able to perfectly recreate the inverted

self-interference signal and therefore cancellation is limited to 25dB

in analog. However this design also uses two separate antennas sep-

arated by 20cm for TX and RX and achieves another 30dB in analog

cancellation via antenna isolation. Hence a total of 55dB of self-

interference reduction is obtained in analog, this cancellation applies

to all the signal components (linear, non-linear and noise). The digi-

tal cancellation stage of this design also only models the linear main

signal component, it does not model the non-linear harmonics that

we discussed above. Thus we found that we obtain another 30dB of

linear cancellation from digital in this design.

Overall, the second design provides 85dB of linear self-interference

cancellation, 55dB of non-linear cancellation and 55dB of analog

noise cancellation. Thus this design falls short of the requirements

by 25dB (especially for the non-linear component). Hence if full du-

plex is enabled over links whose half duplex SNR is 25dB or lower,

then no signal will be decoded. Further to see any throughput im-

provements with full duplex, the half duplex link SNR would have

to be greater than 45dB.

3. OUR DESIGN

In this section we describe the design of our self-interference can-

cellation technique. Our design is a single antenna system (i.e. the

same antenna is used to simultaneously transmit and receive), wide-

band (can handle the widest WiFi bandwidth of 80MHz as well as all

the LTE bandwidths) and truly full duplex (cancels all self-interference

to the receiver noise ﬂoor). The design is a hybrid, i.e., it has both

analog and digital cancellation stages. Note that our hybrid cancel-

lation architecture is not novel, similar architectures have been pro-

posed in prior work [11, 20, 19]. The novelty of our work lies in

the design of the cancellation circuits and algorithms, as well as their

RX&

ﬁxed&&

delays&

Analog&Cancella<on&Circuit&

control&

algorithm&

Digital Cancellation

Eliminates all linear and

non-linear distortion

Σ&

R+aT

R+iT

Circulator&

ADC#

LNA&

variable&

aZen&

Juators&

Σ&

PA&

DAC#

TX&

Figure 3: Full duplex radio block diagram. T

is intended baseband signal

we think we are transmitting, but in fact the transmit signal is T (red). The

intended receive signal is R (green), however we see strong components of

the red signal the RX side. Some of these red signals are undesirably leaked

through the circulator. The analog cancellation circuit is trying to recreate a

signal that matches the leaked interference signal for cancellation. The digital

cancellation stage eliminates any residual self interference.

performance. To the best of our knowledge this is the ﬁrst technique

that achieves 110dB of cancellation and eliminates self-interference

to the noise ﬂoor.

3.1 Analog Cancellation

We introduce a novel analog cancellation circuit and tuning algo-

rithm that robustly provides at least 60dB of self-interference cancel-

lation. Fig. 3 shows the high level design of the circuit and where it

is placed in the radio architecture. A single antenna is connected to

a circulator (at port 2), which is a 3 port device that provides limited

isolation between port 1 and port 3 while letting signals pass through

consecutive ports as seen in Fig. 3. The TX signal is fed through

port 1, which routes it to the antenna connected to port 2, while the

received signal from the antenna is passed from port 2 through to

port 3. Circulator cannot completely isolate port 1 and port 3, so

inevitably the TX signal leaks from port 1 to port 3 and causes inter-

ference to the received signal. From our experiments we ﬁnd that the

circulator only provides 15dB of isolation, i.e., the self-interference

that is leaking to the RX circuit is reduced only by 15dB. To get to

the noise ﬂoor, we still have to provide 95dB of cancellation, and

at least 45 dB of that has to come in analog to ensure transmitter

noise is sufﬁciently canceled and we do not saturate the receiver. We

accomplish this using our novel analog cancellation circuit that we

describe next. Note that when we report analog cancellation perfor-

mance numbers, we include the 15dB of reduction we get from the

circulator for simplicity of description.

Fig. 3 shows the design of our analog cancellation circuit. We

tap the TX chain to obtain a small copy of the transmitted signal

just before it goes to the circulator. This copy therefore includes the

transmitter noise introduced by the TX chain. The copy of the signal

is then passed through a circuit which consists of parallel ﬁxed lines

of varying delays (essentially wires of different lengths) and tunable

attenuators. The lines are then collected back and added up, and this

combined signal is then subtracted from the signal on the receive

path. In effect, the circuit is providing us copies of the transmit-

ted signal delayed by different ﬁxed amounts and programmatically

attenuated by different variable amounts. The key challenge is to

378

评论收藏

内容反馈

ggbasara

2015-06-17

整理的很好实用

咯12345

粉丝: 0
资源: 3

03 SIGCOMM 2013 Session8-12

SIGCOMM 2013-ACM SIGCOMM conference on Data communication 2013论文集

SIGCOMM 2010-ACM SIGCOMM conference on Data communication 2010

02 SIGCOMM 2013 Session1-7

Sigcomm17-Technical Session 1-8

Sigcomm17-Technical Session 9-11 and keynote

SIGCOMM 2012-ACM SIGCOMM conference on Data communication 2012论文集

SIGCOMM 2011-ACM SIGCOMM conference on Data communication 2011论文集

ACM SIGCOMM 2019 Program（2）.7z

SIGCOMM 2017 All Papers

ACM SIGCOMM 2019 Program（1）.7z

SOSP 2013-ACM Symposium on Operating Systems Principles 2013年论文集

IMC 2011-Annual conference on Internet measurement 2011

SOSP 2011-ACM Symposium on Operating Systems Principles 2011年论文集

如何找要读的paper1

2023.10.21 雷蛇+鼠标宏+PUBG+绝地求生 步枪通用

115转存助手ui优化版3.9.1网友魔改-转存提取全修复-user

恩山新版中兴Telnet工具体验版-20230830.rar

【Chrome谷歌浏览器】篡改猴(油猴/tempermongkey)插件安装包

广联达2024最新3226写锁与加密狗授权工具2.6锁

基于华为eNSP的校园网设计和仿真模拟.zip

BPSK基带调制解调的MATLAB实现源码，非库函数

Chrome浏览器 v109.0.5414.75稳定版 64位 离线安装包

蓝牙驱动，解决win11下的设备管理器蓝牙报错提示为：Generic Bluetooth Adapter驱动感叹号解

TVbox（内置源，安装即用）

国家电网DLT645-2007智能电表测试工具

华中科技大学运算器实验源码

大漠插件3.1233全套

广联达盗版环境清理，盗版弹窗清理，广联达防锁

高逼格PPT模板20套.rar

OpenGD77刷机材料

最新资源

2023.10.21 雷蛇+鼠标宏+PUBG+绝地求生步枪通用

Chrome浏览器 v109.0.5414.75稳定版 64位离线安装包