Tolerance to in-Band Crosstalk of Virtual Carrier-assisted Direct

Detection Multi-Band OFDM Systems

Bruno R. Pinheiro

, Jo

ao L. Rebola

1,2

and Adolfo V. T. Cartaxo

1,3

Optical Communications and Photonics, Instituto de Telecomunicac¸

oes, Lisboa, Portugal

Instituto Universit

ario de Lisboa (ISCTE-IUL), Lisboa, Portugal

Instituto Superior Tecnico, Universidade de Lisboa, Lisboa, Portugal

Keywords:

Direct Detection, in-Band Crosstalk, Monte-Carlo Simulation, Multi-Band Orthogonal Frequency Division

Multiplexing.

Abstract:

The tolerance to in-band crosstalk of virtual carrier (VC)-assisted direct detection (DD) multi-band orthogonal

frequency division multiplexing (MB-OFDM) system is assessed numerically through Monte-Carlo simulation

and considering a single interferer. The inﬂuence of the virtual carrier-to-band power ratio (VBPR) and the

virtual carrier-to-band gap (VBG) of the interferer on the in-band tolerance is also studied. We show that,

for interferers with the same VBG as the selected signal, the increase of the VBPR of the interferer leads to

lower optical signal-to-noise ratio (OSNR) penalties. The increase of the VBG of the interferer with central

frequency different from the selected signal also leads to lower OSNR penalties. When the central frequencies

of the interferer and selected bands are the same, the variation VBG of the interferer can lead to 11 dB less

tolerance to in-band crosstalk of the VC-assisted DD OFDM system.

1 INTRODUCTION

Metropolitan networks are responsible for the ag-

gregation of different types of trafﬁc and to pro-

vide a link between access and back-bone networks.

Hence, these networks must present high ﬂexibility,

enabling scalability, dynamic reconﬁgurability and

transparency (Alves, 2015). The virtual carrier (VC)-

assisted direct detection (DD) multi-band orthogonal

frequency-division multiplexing (MB-OFDM) net-

work, proposed in (Alves, 2014), has been appointed

as an efﬁcient approach to provide such requirements

to metro networks (Alves, 2015). This approach is

called MORFEUS (Alves, 2014). The use of a VC

close to each OFDM band enhances the spectral efﬁ-

ciency (SE) and allows the reduction of the required

receiver bandwidth (Peng, 2009).

An important limitation of DD OFDM systems is

the signal-to-signal beat interference (SSBI) caused

by photodetection. In this work, the SSBI mitiga-

tion technique presented in (Nezamalhosseini, 2013)

is implemented, in order to eliminate the SSBI at the

received OFDM signal. The use of the SSBI mitiga-

tion technique allows to reduce the band gap between

the VC and the OFDM band, and consequently, the

system SE is improved (Alves, 2014).

The metro networks performance can be impaired

by in-band crosstalk. The in-band crosstalk is the in-

terference between signals with same nominal wave-

length, and it is originated from the imperfect isola-

tion of the switching devices inside the optical nodes.

This imperfect isolation induces power leakage from

demultiplexed signals on the desired optical signal,

known as crosstalk signals (Winzer, 2011). In-band

crosstalk has been studied in the context of the con-

ventional DD OFDM systems (Rebola, 2014), where

the band gap between the VC and the OFDM band is

equal to the OFDM signal. However, the performance

of a VC-assisted DD OFDM system impaired by in-

band crosstalk is still to be assessed.

In this work, the tolerance to in-band crosstalk of

the VC-assisted DD MB-OFDM system is estimated

through Monte-Carlo (MC) simulation and using the

direct error counting (DEC) as bit error rate (BER) es-

timation method. The error vector magnitude (EVM)

is also used as performance estimation method. We

also evaluate the inﬂuence of the main parameters

of the interferer on the DD OFDM system perfor-

mance, such as the virtual carrier-to-band power ratio

(VBPR) and virtual carrier-to-band gap (VBG) using

the EVM and DEC methods, and the estimates are

compared.

Pinheiro, B., Rebola, J. and Cartaxo, A.

Tolerance to in-Band Crosstalk of Virtual Carrier-assisted Direct Detection Multi-Band OFDM Systems.

DOI: 10.5220/0005743300310037

In Proceedings of the 4th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2016), pages 33-39

ISBN: 978-989-758-174-8

DEMUX

MUX

MIB

OFDM Tx EOC BS

MEB

SEB

OFDM Rx

ROADM of node n

from node n − 1

to node n+1

Node n

MIB MEB

Node 1

MIB MEB

Node 2

MIB MEB

Fiber

Figure 1: Block diagram of the MORFEUS metro network and respective nodes, comprising reconﬁgurable optical add-and-

drop multiplexer (ROADM), MORFEUS insertion block (MIB) and MORFEUS extraction block (MEB). EOC, BS and SEB

stand for electrical-optical converter, band selector and SSBI estimation block, respectively.

This paper is organized as follows. Section 2 de-

scribes the MORFEUS network and its simulation

model. The MORFEUS network is presented in sub-

section 2.1 and, in subsection 2.2, the MC simulation

is described. Numerical results are presented and dis-

cussed in Section 3. Conclusions are outlined in Sec-

tion 4.

2 NETWORK MODEL

In this section, the MORFEUS network and its model

are presented. Then, the main parameters of MB-

OFDM signal are detailed. The MC simulation is also

presented.

2.1 MORFEUS Network

Figure 1 depicts the block diagram of the MOR-

FEUS metro network (Alves, 2015). The MORFEUS

network consists of a ring topology. Each network

node comprises a reconﬁgurable optical add-and-drop

multiplexer (ROADM), a MORFEUS insertion block

(MIB) and a MORFEUS extraction block (MEB). The

MIB is responsible for the generation of the electri-

cal OFDM bands and VCs at the OFDM transmitter

(Tx). Then, the electrical OFDM signal is converted

to the optical domain using an Electrical-to-Optical

Converter (EOC), and inserted in the optical network

(Alves, 2015). In the MEB, the band extraction is per-

formed by a tunable optical ﬁlter (BS), which selects

the desired OFDM signal. Then, the selected OFDM

signal is sent to the SSBI estimation block (SEB) and

also to the PIN photodiode in order to be photode-

tected. Then, the estimated SSBI obtained from the

SEB, is removed from the photodetected OFDM sig-

nal, and after, the signal demodulation is performed at

the OFDM Rx.

The impact of in-band crosstalk on the DD OFDM

receiver performance can be assess considering a sin-

gle OFDM band. Therefore, we assume that the

OFDM signal has only one pair OFDM band-VC,

whose spectrum is depicted in Figure 2.

Figure 2 depicts the OFDM signal at the output

of the transmitter with an average power of 11 mW.

The OFDM band has a bandwidth, B

, of 2.675 GHz

and a central frequency of 5 GHz. The bandwidth B

is deﬁned as N

, where N

is the number of the

subcarriers and T

is OFDM symbol duration without

guard time. The frequency gap between the OFDM

band and the VC is the VBG, which in Figure 2 is

0.5B

. In this work, in order to maximize the system

SE, the VBG is set to 20.9 MHz. The ratio between

the average power of the VC and the power of the cor-

responding OFDM band is the VBPR.

Figure 2: PSD of the electrical MB-OFDM signal at the

OFDM Tx output, with an average power of 11 mW, con-

sidering one pair band-VC.

PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology

(t )

OFDM Tx

with VC

generation

CW Laser

CW Laser DP-MZM

+ +

Crosstalk

ASE noise

OFDM Rx

BER estimation

SEB

Figure 3: System model of the VC-assisted DD OFDM system. DP-MZM, HT and CW stand for dual parallel Mach-Zehnder

modulator, Hilbert transform and continuous wave, respectively.

2.2 Monte Carlo Simulation

In this subsection, the MC simulation and the meth-

ods to assess the BER are described.

Figure 3 depicts the simulation model of the

MORFEUS network considered to assess the toler-

ance of its performance to in-band crosstalk. It con-

sists of an OFDM transmitter (Tx) with VC genera-

tion, a dual parallel Mach-Zehnder modulator (DP-

MZM), a tunable band selector (BS), an ideal pho-

todetector, a SEB and an OFDM receiver. The BER is

estimated at the output of the OFDM receiver. Ampli-

ﬁed spontaneous emission (ASE) noise and in-band

crosstalk are added to the OFDM signal at the optical

receiver input, before the BS.

The MC simulation starts with the generation of

the electrical OFDM signal, comprising the OFDM

band with a VC, and then, the electrical-optical con-

version is performed by the DP-MZM. The DP-MZM

generates a single-side band OFDM optical signal by

applying the electrical OFDM signal and its Hilbert

transform (HT) in the two arms of the modulator. In

this work, the HT of the OFDM signal is considered

ideal. The modulation index of the DP-MZM is set to

the optimized value of 5% (Alves, 2015). Then, by as-

suming a back-to-back conﬁguration, ASE noise and

in-band crosstalk sample functions are added to the

optical OFDM signal.

The model of the SEB is depicted in Figure 4, and

it is based on the SSBI mitigation technique presented

in (Nezamalhosseini, 2013). The SEB is composed of

two branches. At the lower branch, the VC of the se-

lected OFDM signal is selected using an ideal optical

ﬁlter, named virtual carrier selector (VCS), and then,

the VC is removed from the OFDM signal in the up-

VCS

Figure 4: Model of the SEB. VCS stands for VC selector.

per branch. Afterwards, the SSBI is estimated after

the photodetection of the OFDM signal without the

VC. To conclude the SSBI mitigation algorithm, the

SSBI is removed from the photodetected OFDM sig-

nal before arriving the OFDM receiver, as shown in

Figure 3. In this work, we assume that both branches

of the SEB are synchronized.

At the BS input, the OFDM signal, s

(t), impaired

by the interferer and ASE noise can be written as

(t) = s

(t) +

∑

i=1

x,i

(t −τ

jφ

+ N

(t) (1)

where s

(t) is the selected OFDM signal, s

x,i

(t) is the

i-th interfering signal of N

interferers and N

(t) is the

complex envelope of the ASE noise. We assume that,

the ASE noise follows a zero mean Gaussian distri-

bution with variance of N

sim

, where N

is the ASE

noise power spectrum density and B

sim

is the band-

width used in the MC simulation. τ

and φ

are, re-

spectively, the time delay and the phase difference be-

tween the selected and the i-th interfering signals. τ

is modeled as a uniformly distributed random variable

between zero and T

, and φ

has a uniform distribu-

tion within the interval [0, 2π] (Winzer, 2011). The

relation between the average powers of the i-th inter-

ferer and the selected OFDM signal is deﬁned as the

crosstalk level (Winzer, 2011). In each iteration of the

MC simulation, a sample function of ASE noise and

of in-band crosstalk are generated and added to the

optical OFDM signal.

When estimating the EVM, the MC simulation

stops after 75 iterations (Alves, 2010), and then, the

root mean square (rms) of the EVM of each OFDM

subcarrier is evaluated using (Alves, 2010)

EV M

rms

[k] =

∑

n=1

[k] −s

[k]|

∑

n=1

[k]|

k ∈ {1, 2,..., N

}

(2)

where s

[k] and s

[k] are, respectively, the received

and the transmitted symbol at the k -th subcarrier of

each n-th OFDM symbol of the total number of gen-

erated OFDM symbols, N

. Then, the BER of each

subcarrier, BER[k] is computed from (Shaﬁk, 2006)

Tolerance to in-Band Crosstalk of Virtual Carrier-assisted Direct Detection Multi-Band OFDM Systems

BER[k] = 4

(1 −1/

√

log

(M)

(M −1) ·EV M

rms

[k]

(3)

and the overall BER of the OFDM signal is given by

BER =

∑

k=1

BER[k] (4)

Remark that Equation 3 assumes a Gaussian distribu-

tion for the distortion each subcarrier (Alves, 2010).

The BER is estimated from DEC after a total of

5000 counted errors, N

, is reached in the OFDM re-

ceived signal, and is obtained using N

/(N

where N

is the number of iterations of the MC simu-

lation and N

is the number of bits per symbol in each

OFDM subcarrier (Alves, 2010).

3 RESULTS AND DISCUSSION

In this section, the tolerance to in-band crosstalk of

the VC-assisted DD MB-OFDM communication sys-

tem is assessed numerically. The tolerated crosstalk

level, X

c,max

, is deﬁned as the crosstalk level that

leads to a 1 dB optical signal-to-noise ratio (OSNR)

penalty. The OSNR penalty is deﬁned as the dif-

ference in dB between the OSNR in presence of

crosstalk and the OSNR without crosstalk that lead

to a BER of 10

−3

(Winzer, 2011).

Table 1 presents the parameters used in MC sim-

ulation, in order to assess the tolerance to in-band

crosstalk of the VC-assisted DD OFDM system. The

VBPR, the −3 dB bandwidth of the BS, which in this

work is a 2

-order Super-Gaussian, and the modu-

lation index are obtained from the optimization per-

formed in (Alves, 2015). The parameters of the se-

lected OFDM signal are kept the same throughout

this work. The parameters of the interferer are equal

Table 1: Simulation parameters of the VC-assisted DD

OFDM system.

Bit rate per band [Gbps] 10.7

Number of subcarriers (N

) 128

Bandwidth [GHz] 2.675

OFDM symbol duration [ns] 47.85

Central frequency [GHz] 5

VBPR [dB] 6

VBG [MHz] 20.9

-3 dB bandwidth of BS [GHz] 3.6

modulation index 5%

OSNR @BER=10

−3

[dB] 15.3

modulation format 16-QAM

Crosstalk level [dB]

-35 -34 -33 -32 -31 -30 -29 -28 -27 -26 -25 -24 -23 -22

OSNR penalty [dB]

0.5

1.5

2.5

VBPR=0 dB

VBPR=2 dB

VBPR=4 dB

VBPR=6 dB

VBPR=8 dB

VBPR=10 dB

VBPR=12 dB

Figure 5: OSNR penalty as a function of the crosstalk level

obtained from the EVM method, for interfering signals with

different VBPRs.

to the ones of the selected OFDM signal, except for

the VBPR and VBG. The OSNR for a BER of 10

−3

without crosstalk is obtained using the EVM, and is

in agreement with the OSNR obtained using DEC.

In subsection 3.1, the VBPR of the interferer is

changed and its inﬂuence on the VC-assisted DD

OFDM system performance is assessed. In subsec-

tion 3.2, the impact of the VBG of the interferer on the

in-band crosstalk tolerance is evaluated considering

two distinct scenarios: in scenario (a), the frequen-

cies of the selected signal and interferer VCs are the

same, and in scenario (b) the central frequencies of

the OFDM band of the selected signal and interferers

are equal.

3.1 Inﬂuence of the VBPR on the

in-Band Tolerance

Figure 5 depicts the OSNR penalty as a function of

the crosstalk level due to a single interferer having

different VBPRs than the selected OFDM signal, ob-

tained using the EVM method. Figure 5 shows that

higher VBPRs of the interferer lead to lower OSNR

penalties. In Figure 6, the tolerated crosstalk level,

obtained from Figure 5, is depicted as a function of

the VBPR with a solid line. Figure 6 shows also the

tolerated crosstalk level as a function of the VBPR,

estimated using DEC, with a dashed line. A differ-

ence of about 2.8 dB between the tolerated crosstalk

level for VBPR of 0 dB and 12 dB is observed,

and the receiver performance degradation enhances

with the increase of the VBPR of the interferer. In

this case, the interfering band overlaps the selected

OFDM band in the frequency domain, and therefore,

the increase of the VBPR of the interferer reduces the

power of its OFDM band, hence, leading to less inter-

ference.

The tolerated crosstalk levels obtained using the

PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology

VBPR [dB]

0 2 4 6 8 10 12

c,max

[dB]

-29

-28

-27

-26

-25

-24

-23

EVM estimation

DEC estimation

Figure 6: Tolerated crosstalk level as a function of the

VBPR of the interferer, considering DEC (dashed lines) and

EVM (solid lines) estimations.

DEC method are 0.8 to 1.8 dB higher than the ones

obtained using EVM method, and the difference is

higher with the increase of the VBPR of the interferer.

As an example, for a VBPR of 6 dB, the EVM esti-

mates a tolerated crosstalk level of −26.7 dB, while,

the DEC predicts a tolerated crosstalk of −25.6 dB.

Considering these discrepancies between both meth-

ods, we conclude that the EVM is inaccurate on the

tolerated crosstalk level estimation, when the selected

and interferer OFDM signal have the same VBG but

different VBPRs. The in-band crosstalk sample func-

tions are not modelled by a Gaussian distribution,

thus, the BER obtained from Equation 3 can lead to

poor BER accuracy (Alves, 2010).

3.2 Inﬂuence of the VBG on the

in-Band Tolerance

In this subsection, the impact of the VBG of the in-

terferer on the VC-assisted DD OFDM system per-

formance is assessed, considering two different sce-

narios. In scenario (a), the frequency of the VC of

the interfering OFDM signal is the same as the VC of

the selected signal, and the variation of the VBG of

the interferer leads to a frequency deviation between

the central frequencies of the selected and interferer

OFDM bands, as it can be seen in Figure 7(a). In Fig-

ure 7(a), the spectrum of the selected OFDM signal

is depicted in black and the interfering OFDM sig-

nal spectrum is shown in gray. The interferer has a

crosstalk level of −20 dB and a VBG of 1.34 GHz.

Figure 7(b) exempliﬁes the scenario (b), in which, the

central frequencies of the interfering and the selected

OFDM bands are the same, and the variation of the

VBG of the interferer leads to a frequency difference

between the VC frequencies of the selected and inter-

ferer signals. In Figure 7(b), the VBG of the interferer

(a)

(b)

Figure 7: PSDs of the selected OFDM signal (black) and

interferer signal (gray) with a crosstalk level of −20 dB and

a VBG=B

/2 considering (a) same VCs frequencies and (b)

equal OFDM bands central frequencies.

is also 1.34 GHz.

Figure 8 depicts the tolerated crosstalk level as a

function of the VBG of the interferer, for both sim-

ulation scenarios. Focusing on scenario (a), Figure 8

shows that the tolerance to in-band crosstalk increases

with the VBG of the interferer. The crosstalk level

VBG [GHz]

0,02 0.33 0.67 1 1.34 1.62 2 2.34 2.68

c,max

[dB]

-38

-35

-32

-29

-26

-23

-20

-17

-14

Scenario (b) EVM

Scenario (b) DEC

Scenario (a) EVM

Scenario (a) DEC

Figure 8: Tolerated crosstalk level as a function of the VBG

considering the DEC (dashed lines) and EVM (solid lines)

estimations.

Tolerance to in-Band Crosstalk of Virtual Carrier-assisted Direct Detection Multi-Band OFDM Systems

with a VBG of 2.34 GHz is about 9 dB higher than the

one obtained with a VBG of 20.9 MHz. As already

said, this scenario leads to the misalignment between

the central frequencies of the interferer and selected

bands, thereby, as the VBG increases, less subcarriers

of the selected band are affected by in-band crosstalk,

and consequently, the robustness of the DD OFDM

system to in-band crosstalk is enhanced. For a VBG

of 2.68 GHz, the interferer OFDM band is totally mis-

aligned from the selected signal OFDM band, which

means, that the OFDM subcarriers of the selected sig-

nal are not affected by in-band crosstalk, leading to a

null OSNR penalty. Hence, we can conclude that, in

scenario (a), the DD OFDM receiver is completely

tolerant to interfering OFDM signals with VBG equal

or wider than the selected OFDM signal bandwidth.

Figure 8 shows that the EVM estimations for

the tolerated crosstalk level are in disagreement with

the DEC estimations, since a difference below 2 dB

between both estimations is observed. Once again,

this disagreement between the two methods is at-

tributed to the non-Gaussian distribution of the in-

band crosstalk sample functions.

Regarding scenario (b), Figure 8 shows that in-

terfering signals with a VBG between 83.6 MHz and

0.67 GHz exhibit a signiﬁcant reduction on the toler-

ated crosstalk level, in comparison with smaller VBG.

For a VBG of 83.6 MHz, the tolerated crosstalk level

is about 11 dB lower than the one obtained for a VBG

of 20.9 MHz.

In order to investigate this behavior, the BER es-

timated using the EVM as a function of the subcar-

rier index is depicted in Figure 9. In Figure 9, the

crosstalk level is set to −26 dB, in order to enable

a performance comparison with different VBGs of

the interferer. From Figure 9, it is clear that for a

VBG of 83.6 MHz, the subcarrier 128 is the sub-

carrier with the worst performance due to the pres-

ence of the detected VC interferer. The increase of

Subcarrier index

1 16 32 48 64 80 96 112 128

log

(BER)

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

VBG=20.9 MHz

VBG=83.6 MHz

VBG=0.34 GHz

VBG=1.34 GHz

Figure 9: BER as a function of the subcarrier index.

(a)

(b)

Figure 10: PSDs of the photodetected signal for a crosstalk

level of −20 dB and (a) VBG of 83.6 MHz and (b) VBG of

1.34 GHz on the interfering OFDM signal.

the VBG of the interferer changes the subcarrier with

worst performance and the BER per subcarrier is re-

duced, hence, decreasing the overall BER. It was ver-

iﬁed that this effect is due to the BS ﬁltering. As the

VBG increases, the frequency of the interferer VC be-

comes closer to the cut-off region of the BS, hence,

the power of the VC is attenuated, and after photode-

tection, the performance of the subcarrier that suffers

the VC interference is improved. For interferers with

a VBG of 2.68 GHz, the frequency of interferer VC is

outside the BS passband, leading to a fully suppres-

sion of the interfering VC. This can be inferred from

the comparison of the OFDM signal spectrums at the

photodetector output depicted in Figure 10.

Figure 10(a) shows the photodetection of the se-

lected band in presence of an interfering band having

a VBG of 83.6 MHz, while, in Figure 10(b), the VBG

of the interferer band is 1.34 GHz. By comparing both

ﬁgures, it is clear that the increase of the VBG leads

to a frequency shift of the detected interferer VC and

to an attenuation of its power.

Comparing the tolerated crosstalk levels estimated

from both methods, in scenario (b), Figure 8 shows

that, for interferers with VBGs between 83.6 MHz

and 0.67 GHz, the tolerated crosstalk level estima-

PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology

tions obtained from EVM and DEC are in agree-

ment. For the remaining VBGs, in which higher tol-

erated crosstalk levels are estimated, a maximum dif-

ference of 1.2 dB between both estimations is noticed.

Hence, we can conclude that the increase of the toler-

ated crosstalk level leads to discrepancies between the

EVM and the DEC estimations, as the BER estimated

from EVM, using Equation 3, looses accuracy.

4 CONCLUSIONS

In this work, the tolerance to in-band crosstalk of

the VC-assisted DD MB-OFDM system has been as-

sessed numerically. The inﬂuence of the VBPR and

VBG of the interferer on the in-band crosstalk toler-

ance has been analyzed.

Our results show that the impact of in-band

crosstalk on the DD OFDM system performance di-

minishes with the increase of the VBPR of the inter-

ferer. The increase of the power of the interferer VC

leads to a power reduction of the interferer OFDM

band, hence, causing less interference on the selected

signal.

The inﬂuence of the VBG of the interfering sig-

nal on the in-band tolerance has been evaluated con-

sidering two different scenarios. In scenario (a), the

frequency of the VCs of the selected signal and inter-

ferer are equal. In this case, the tolerance to in-band

crosstalk reduces with the amount of superposition of

interferer and selected OFDM bands. Larger VBG

leads to a reduction of the number of subcarriers that

are affected by crosstalk and the system performance

improves. When the VBG is equal to the OFDM sig-

nal bandwidth, the in-band crosstalk has no inﬂuence

on the DD OFDM receiver performance degradation.

In scenario (b), the central frequencies of the se-

lected signal and interferer bands are equal. In this

case, the tolerance to in-band crosstalk is severely di-

minished for a VBG of 83.6 MHz, leading to a tol-

erated crosstalk level that is about 11 dB lower that

the ones obtained for very narrow VBGs. This effect

is due to the detection of the interferer VC on the se-

lected signal and leads to a strong performance degra-

dation. As the VBG of the interferer becomes larger,

the BS ﬁltering reduces the power of the VC of the

interferer and the performance of the DD OFDM re-

ceiver is improved.

The comparison between the numerical results ob-

tained from the DEC and EVM methods shows that,

for higher crosstalk levels, a maximum difference of

2 dB between both methods estimations has been no-

ticed. When the tolerated crosstalk level is lower

than −30 dB, the EVM and the DEC estimations are

in agreement. Hence, we can conclude that higher

crosstalk levels lead to an inaccurate BER estima-

tions.

ACKNOWLEDGEMENTS

This work was supported by Fundac¸

ao para a Ci

encia

e Tecnologia (FCT) from Portugal through Project

MORFEUS-PTDC/EEITEL/2573/2012. The project

UID/EEA/50008/2013 is also acknowledged.

REFERENCES

Alves, T., Alberto, A., and Cartaxo, A. (2014). Direct-

Detection Multi-Band OFDM Metro Networks Em-

ploying Virtual Carriers and Low Receiver Band-

width. Opt. Fiber Commun. Conf.

Alves, T. and Cartaxo, A. (2015). High Granularity Multi-

band OFDM Virtual Carrier-assisted Direct-Detection

Metro Networks. 33(1):42–54, paper Tu3G.5.

Alves, T. and Cartaxo, A. (2010). Analysis of Methods of

Performance Evaluation of Direct-Detection OFDM

Communication Systems. Fiber and Integrated Op-

tics, 29(3):170 – 186.

Nezamalhosseini, S., Chen, L., Zhuge, Q., Malekiha, M.,

Marvasti, F., and Plant, D. (2013). Theoretical and Ex-

perimental Investigation of Direct Detection Optical

OFDM Transmission using Beat Interference Cancel-

lation Receiver. Opt. Express, 21(13):15237–15246.

Peng, W., Zhang, B., Feng, K., Wu, X., Willner, A., and Chi,

S. (2009). Spectrally Efﬁcient Direct-Detected OFDM

Transmission Incorporating a Tunable Frequency Gap

and an Iterative Detection Techniques. J. Light. Tech-

nol., 27(24):5723–5735.

Rebola, J. and Cartaxo, A. (2014). Impact of In-Band

Crosstalk Due to Mixed Modulation Formats with

Multiple Line Rates on Direct-Detection OFDM Opti-

cal Networks Performance. In Int. Conf. Transparent

Opt. Networks, pages 14–17.

Shaﬁk, R., Rahman, S., and Islam, R. (2006). On the Ex-

tended Relationships Among EVM, BER and SNR as

Performance Metrics. In International Conference on

Electrical and Computer Engineering (ICECE), pages

408–411.

Winzer, P., Gnauck, A., Konczykowska, A., Jorge, F., and

Dupuy, J. (2011). Penalties from In-Band Crosstalk

for Advanced Optical Modulation Formats. In Euro-

pean Conference and Exhibition on Optical Commu-

nication (ECOC), paper Tu.5.B.7.

Tolerance to in-Band Crosstalk of Virtual Carrier-assisted Direct Detection Multi-Band OFDM Systems