power measurements in dvb-t systems: new proposal...
TRANSCRIPT
IM-6241
POWER MEASUREMENTS IN DVB-T SYSTEMS:
NEW PROPOSAL FOR ENHANCING RELIABILITY AND REPEATABILITY
Leopoldo Angrisani, Domenico Capriglione, Member, IEEE,
Luigi Ferrigno, Member, IEEE, and Gianfranco Miele, Student Member, IEEE
Leopoldo Angrisani DIS University of Naples “Federico II” Via Claudio 21 80125 Napoli, Italy Phone:(39) 081-7683170 Fax:(39) 081-7683816 E_mail: [email protected] Domenico Capriglione DAEIMI University of Cassino Via G.Di Biasio 43 03043 Cassino (FR), Italy Phone:(39) 0776-2993673 Fax: (39) 0776-2993707 E_mail: [email protected] Luigi Ferrigno DAEIMI University of Cassino Via G.Di Biasio 43 03043 Cassino (FR), Italy Phone:(39) 0776-2993672 Fax: (39) 0776-2993707 E_mail: [email protected]
Gianfranco Miele DAEIMI University of Cassino Via G.Di Biasio 43 03043 Cassino (FR), Italy Phone:(39) 0776-2993683 Fax: (39) 0776-2993707 E_mail: [email protected]
IM-6241
Abstract – The rapid growth of digital video broadcasting by terrestrial transmission (DVB-T) has
created the need for getting new measurement methods as well as new test equipment up and capable of
providing reliable and repeatable results. Some problems are, in fact, being experienced in power
measurements, especially in those involving the integration of input signal power spectrum over a certain
frequency interval.
Trying to give a proper answer to the cited need, a digital signal processing method, already proposed in
literature and specialized for power measurements in spread spectrum systems, is here optimized to be
operative also in the presence of DVB-T signals. Suitable stages, in simulation and emulation
environment, are designed and applied in order to optimally regulate key parameters of the method, thus
making it assure negligible bias and high repeatability in most operative conditions. The results of a
number of experiments on actual DVB-T signals give also evidence of the method’s efficacy with respect
to competitive measurement solutions.
Keywords – Power measurement, Channel power measurement, RF measurements, DVB-T, WOSA
estimation, Multitaper estimation, Power spectrum estimation, Spectrum analyzer, VSA, RTSA.
I. INTRODUCTION
A new era in television broadcasting has dawned with the introduction of DVB (Digital Video
Broadcasting) standard. Whether via satellite (DVB-S), cable (DVB-C) or by terrestrial transmission
(DVB-T) to existing antennas, DVB is revolutionizing television transmission. It represents a
considerable technological opportunity that allows television broadcasters to provide services meeting the
IM-6241
expectations of more and more demanding and advanced users, such as the ones of the twenty-first
century.
With special regard to DVB-T, many advantages of digital television over analog one can be enlisted: (i)
more than one program in the same occupied radiofrequency bandwidth (typically four or more), (ii)
lower radiofrequency (RF) power required to cover the same distance (i.e. greater immunity to noise and
disturbances), (iii) better picture quality, (iv) possibility of building single-frequency networks (SFN’s),
(v) mobile reception, completely precluded in analog systems, and (vi) availability of interactive MHP
(Multimedia Home Platform) applications [1].
A new set of measurements for assessing the performance of DVB-T systems and apparatuses is also
required [2]. Due to the composite modulation technique adopted (COFDM, Coded Orthogonal
Frequency Division Multiplexing), a completely new measurement challenge is issued for RF signal
integrity and physical layer analysis. COFDM provides for a high-density, multi-carrier scheme, which
allows the construction of large SFN’s and helps to mitigate multipath problems in a crowded
environment. Digital data is, in fact, conveyed on 1705 (2k transmission mode) or 6817 (8k transmission
mode) carriers within the 8 MHz bandwidth of a traditional analogue channel.
As highlighted in Tab. I, power measurements play a very important role. RF and IF (intermediate
frequency) signal power, RF channel power, RF and IF power spectrum, noise (or unwanted) power, and
power efficiency are relevant parameters to be measured as accurately as possible [2]. To this aim, RF
power meters equipped with proper probes and spectrum analyzers are available on the market. The formers
are specifically mandated to peak and average power measurements, while the use of the latter is mandatory
whenever the integration of input signal power spectrum over a certain frequency range (for example, in
channel power measurement) is involved [3]. Spectrum analyzers, however, suffer from low accuracy and
repeatability problems, the entity of which depends on the specific type of power measurement. This is
mainly due to the high PAR (peak-to-average power ratio), and consequently high crest factor,
IM-6241
characterizing COFDM signals. Peak power can be much greater than average one, thus making the signal
exhibit rapid changes in time domain (white noise-like nature) that can disturb the proper operation of the
cited instruments [4]. Measurement strategies suggested to mitigate these problems assure negligible
advantages if compared to the long time their application require [5].
One of the authors proposed in [6] a new method for power measurement in digital wireless
communication systems, expressly tailored to spread spectrum modulation schemes. The method mainly
aimed at: (i) simultaneously and automatically carrying out average and channel power measurements, (ii)
overcoming the limits of spectrum analyzers, thus providing much more reliable and repeatable results,
and (iii) granting as good measuring accuracy as that guaranteed by the latest generation of power meters.
After down-converting and digitizing the input RF signal, the method first estimates its true power
spectrum using well-known digital signal-processing solutions, such as weighted overlapped segment
averaging (WOSA) [7] and multitaper [8],[9] estimators and, then, evaluates the quantities of interest
through very straightforward measurement algorithms.
The paper aims at optimizing the aforementioned method to make it operate with success also in DVB-T
systems. Simulation and emulation stages are properly designed in order to regulate the most relevant
parameters of the method according to the specific features of signals involved; signal conditions very
close to real ones are, in particular, induced. A number of experiments on actual DVB-T signals are also
conducted. The results peculiar both to the emulation stage and actual signals are compared to those
provided by competitive measurement solutions, such as high-performance power meters, traditional
spectrum analyzers, vector signal analyzers and real time spectrum analyzers.
IM-6241
II. MEASUREMENT METHOD
The measurement method adopted by the authors is roughly sketched in Fig. 1. It can be subdivided into
four sections, the description of which, suitably complemented with theoretical remarks, is given in the
following. Further details can be found in [6].
A. Conditioning Section
The incoming RF signal can be down-converted to a suitable intermediate frequency (IF) or not according
to the bandwidth, maximum sample rate, and memory depth of the available data acquisition system.
B. Digitizing Section
RF or IF input signal has to be digitized. The acquired record is stored and passed to the PSD estimation
section.
C. PSD Estimation Section
A proper digital signal processing algorithm is applied to the acquired samples in order to estimate the
power spectral density (PSD) of the DVB-T signal. Two major algorithms for PSD estimation have been
taken in account [6]. The first is based on the Welch method of averaged periodograms, also known as
WOSA estimator; the second applies wavelet thresholding techniques to the logarithm of the multitaper
estimator.
WOSA Estimator: The WOSA estimator is computationally one of the most efficient methods of PSD
estimation, especially for long data records [7]. This method is based on the division of the acquired
signal, x(n), into smaller units called segments, which may overlap or be disjoint. The samples in a
IM-6241
segment are weighted through a window function in order to reduce undesirable effects related to spectral
leakage. For each segment, a periodogram
( ) ( ) ( )21
2
0
SS
Nj fnTi iS
xnS
TS f x n n eN U
πω−
−
=
= ∑ (1)
is calculated. The variable f stands for frequency, xi(n) are the samples of the i-th segment, w(n) accounts
for the window coefficients, NS denotes the number of samples in a segment, U is a coefficient given by
( )
12
0
1 SN
nS
U nN
ω−
=
= ∑ (2)
and used to remove the window effect from the total signal power, and TS represents the sampling period.
The PSD estimate Sx(f) is then computed by averaging the periodogram estimates
( ) ( )1
0
1 Ki
xi
S f S fK
−
=
= ∑ x (3)
where K represents the number of segments and is given by
1S
S P
N NKN N
−= +
− (4)
N stands for the total number of acquired samples, and NP is the number of the overlapped samples
between two successive segments. The overlap ratio r is defined as the percentage of ratio between the
number of the overlapped samples and the number of samples in a segment
100 %P
S
NrN
= (5)
It is worth noting that a proper use of the WOSA estimator imposes the optimal choice of two parameters:
the window function ω(·) and the overlap ratio r. Periodogram in (1) can be easily evaluated over a grid of
equally spaced frequencies through a standard Fast Fourier Transform (FFT) algorithm [7].
IM-6241
Multitaper Estimation and Wavelet thresholding: The idea is to calculate a certain number, H, of PSD
estimates, each using a different window function, also called data taper and applied to the whole acquired
signal, and then to average them together [8]. If all data tapers are orthogonal, the resulting multitaper
estimator can exhibit good performance, in terms of reduced bias and variance, particularly for signals
characterized by high dynamic range and/or rapid variations like those peculiar to DVB-T systems.
The multitaper estimator has the following form
( ) ( )
1
0
1 Hi
x xi
S f S fH
−
=
= ∑ (6)
where the terms Six( f ), called eigenspectra, are given by
( ) ( ) ( )
12
0
S
Nj fnTi
x in
S f x n h n e π−
−
=
= ∑ (7)
{hi(n): n = 0, …, N-1; i = 1, …, H} denotes a set of orthonormal data tapers. A convenient set of easily
computable orthonormal data tapers is the set of sine tapers, the ith of which is
( ) ( )1/ 2 12 sin
1i
i nh n
N N 1π+⎛ ⎞⎛ ⎞= ⎜ ⎟⎜ ⎟+⎝ ⎠ ⎝ + ⎠ (8)
A standard FFT algorithm proves appropriate to evaluate the eigenspectra over a grid of equally spaced
frequencies [9].
Provided H equal or greater than 5, it can be demonstrated that the random variable η( f )
( ) ( )
( ) ( )log logxS ff H H
S fη ψ= − +
(9)
has Gaussian distribution with zero mean and variance σ2η equal to ψ´(H); S( f ) represents the true PSD,
ψ(·) and ψ´(·) denote, respectively the digamma and trigamma functions [8]. If we let
( ) ( ) ( )log logxY f S f H Hψ= − + (10)
IM-6241
we have
( ) ( ) ( )logY f S f fη= + (11)
i.e., the logarithm of the multitaper estimator, plus a known constant, can be written as the true log-
spectrum plus approximately Gaussian noise with zero mean value and known variance σ2η
These conditions make wavelet thresholding techniques particularly suitable to remove noise and, thus, to
produce a smooth estimate of the logarithm of the PSD. In particular, after evaluating the discrete wavelet
transform (DWT) of Y( f ), computed according to (10), the resulting wavelet coefficients, which are also
Gaussian distributed, can be subjected to a thresholding procedure, and the aforementioned smooth
estimate can be obtained by applying the inverse DWT to the thresholded coefficients [9]. A soft
threshold function, δ(α,T), is suggested, and it is defined by:
( ) ( ) ,
, sgn0,
T if TT
otherwiseα α
δ α α⎧ − >
= ⎨⎩ (12)
where α denotes the generic wavelet coefficient and T is the threshold level. In [10] Donoho and
Johnstone demonstrated that, in the presence of Gaussian noise with zero mean value and variance σ2η,
the optimal value of T is
2 logT ησ= ⋅ N (13)
where N, the number of samples, must be power of two.
Also in this case, a right choice of two parameters, the number of data tapers H and the mother wavelet
ζ(·) for DWT and inverse DWT evaluation, has to be made for gaining a sound spectral estimation.
D. Power Measurement Section
Once the PSD of the analyzed DVB-T signal has been estimated, each measurement can be carried out by
using very straightforward algorithms [6].
IM-6241
1) With regard to channel power, the tune frequency of the monitored DVB-T channel is first
established. Then, the desired power is obtained by integrating the PSD over the frequency
interval centered at the tune frequency and as wide as the nominal spacing of the channel itself [1].
2) To evaluate signal power (total power), the PSD is integrated over the whole frequency span
analyzed, from zero up to the half of the adopted sample rate fS (fS = 1/Ts).
III. MEASUREMENT METHOD OPTIMIZATION
To optimally choose the window function ω(·) and overlap ratio r for the WOSA estimator and the
number of data tapers H and mother wavelet ζ(·) for the multitaper estimator, a suitable simulation stage
has been designed. For each parameter, the same values considered in [6] have been enlisted.
A number of numerical tests have, in particular, been executed in Matlab 7TM environment with the aim of
minimizing the following figures of merit:
a) experimental standard deviation characterizing both total (σΤ) and channel (σC) power
measurement results;
b) difference between the mean value of the results provided by the method and imposed one,
considered as reference, for both total (∆Τ) and channel (∆C) power.
DVB-T reference signals have firstly been generated. To this aim, the analytical expression for the PSD of
a DVB-T signal, given by
( ) ( )( )( )( )( )( )
( ) 21 / 2
1 / 2
sinKk u
X kk K k u
f f T kS f f fcuf f T
ππ
−
=− −
⎡ ⎤− ∆ += ⎢ ⎥
− ∆ +⎢ ⎥⎣ ⎦∑ T
= + (14)
IM-6241
has been considered [1]; fc is the RF signal central frequency, K the number of transmitted carriers, ∆ the
duration of the guard interval and Tu the reciprocal of carrier spacing. Moreover, the approximate method
in frequency-domain presented in [11] has been adopted. It assures accurate time-domain realizations of a
zero mean Gaussian process, characterized by a known PSD.
The following DVB-T transmission settings have been imposed: 8k transmission mode (K=6817 and
Tu=896 µs), 1/4 (∆=224 µs) and 1/32 (∆=28 µs) guard intervals. In addition, three values of the
oversampling factor (considered as the ratio between the sample rate and RF signal central frequency)
have been simulated, and the hypothesis of acquired records covering one DVB-T symbol has been hold.
For each transmission setting and oversampling factor value, 50 different realizations (test signals) have
been produced.
Obtained results are given in Tab. II and Tab. III for the multitaper and WOSA estimator, respectively.
Each pair of round brackets describes the couple (ζ(·)-H or ω(·)-r) that minimizes the related figure of
merit. The last row of both tables quantifies the computation burden in terms of mean processing time on
a common Pentium IVTM computer.
From the analysis of the results some considerations can be drawn:
• both estimators have assured good repeatability; experimental standard deviation is always lower
than 0.20%;
• repeatability improves upon the widening of the guard interval, and the oversampling factor seems
to have no influence;
• WOSA estimator exhibits better performance in terms of ∆Τ and ∆C;
• measurement time peculiar to the multitaper estimator is much longer than that taken by WOSA
estimator;
IM-6241
• with regard to the WOSA estimator, Ollila window combined with an overlap ratio in the interval
70–90% shows itself the most adequate to assure acceptable results in any measurement;
• concerning the multitaper estimator, a number of tapers equal to 25 and db3 as mother wavelet can
reveal the most convenient choice to assure satisfactory values to each figure of merit.
WOSA estimator has given the better trade-off between metrological performance and measurement time,
thus confirming the outcomes presented in [12]. This is the reason why the multitaper estimator has no
longer been considered in the subsequent stages of the work.
To fix the minimum hardware requirements of the DAS (Data Acquisition System) to be adopted in the
experiments on emulated and actual DVB-T signals described in the following sections, further tests have
been carried out. Sensitivity of the proposed method to ENOB (Effective Number Of Bits) and acquired
record length has been assessed. Ollila window and an overlap ratio of 80% have been taken on. Obtained
results are given in Fig. 2 and Fig. 3; they refer to a guard interval equal to 224 µs. In particular, Fig. 2
shows the values of σC (Fig.2a), ∆C (Fig.2b) and ∆T (Fig.2c) versus ENOB for three values of the
oversampling factor; Fig. 2d presents the estimated PSD for the considered values of ENOB. With regard
to σT, values very similar to those characterizing σC have been experienced. Fig. 3 shows the values of
σΤ (Fig. 3a) and ∆T (Fig. 3b) versus acquired record length, for the same values of the oversampling
factor. With regard to σC and ∆C, values very similar to those characterizing respectively σT and ∆T have
been experienced.
Looking at Fig. 2, it is possible to establish that: (i) ENOB equal or greater than six grants an
experimental standard deviation both in total (σΤ) and channel (σC) power measurement less than 0.15%;
(ii) ∆C does not seem to be affected by vertical quantization, as, on the contrary, ∆T does. Furthermore,
Fig. 3 clearly evidences that σΤ improves upon the widening of the record length, while satisfying values
of ∆Τ can be achieved if record lengths covering more than 1/2 DVB-T symbol are considered.
IM-6241
These considerations match well with the typical characteristics of the data acquisition systems available
today on the market. High sample rates, needed to optimally acquire RF or IF DVB-T signals, are often
associated with ENOB not lower than 6 bits.
IV. PERFORMANCE ASSESSMENT
An emulation stage has been designed and applied with the aim of assessing the performance of the
optimized method in the presence of a real DAS, and comparing it to that assured by competitive
measurement solutions already available on the market. Stemming from the past experience documented
in [12], a suitable measurement station, sketched in Fig. 4, has been set-up. It has included: (i) a
processing and control unit, namely a personal computer (PC), on which the measurement algorithm has
run; (ii) a RF signal generator equipped with DVB-T personalities, Agilent TechnologiesTM E4438C
(250 kHz-6 GHz output frequency range); (iii) a traditional spectrum analyzer (ESA) Agilent
TechnologiesTM E4402B (9 kHz-3 GHz input frequency range); (iv) a vector signal analyzer (VSA)
Agilent TechnologiesTM E4406A (7 MHz-4 GHz input frequency range); (v) a real-time spectrum analyzer
(RTSA) TektronixTM RSA3408A (DC-8 GHz input frequency range); (vi) a RF power meter Agilent
TechnologiesTM N1911A, equipped with two probes N1921ATM (50 MHz-18 GHz input frequency range)
and E9304ATM (6 kHz-6 GHz input frequency range); (vii) a DAS LeCroyTM SDA6000A, (6 GHz
bandwidth, 20 GS/s maximum sample rate). They are all interconnected through an IEEE-488 interface
bus. The function generator has provided 8MHz bandwidth, DVB-T test signals, characterized by a RF
central frequency equal to 610 MHz, a nominal total power of -20 dBm, and a 64-QAM modulation
scheme. Moreover, the same transmission settings considered in the previous stage have been imposed.
IM-6241
A preliminary characterization of cables and connectors utilized in the measurement station has been
carried out through the vector network analyzer ANRITSU 37347CTM (40 MHz-20 GHz input frequency
range), equipped with 3650 SMA 3.5 mm calibration kit [13]. Mean value and experimental standard
deviation of 100 attenuation measures obtained in the interval 606-614 MHz are given in Tab. IV.
Different operative conditions of the DAS, in terms of vertical resolution (7 and 8 bit nominal) and
observation period (1/4, 1/2, 3/4 and 1 DVB-T symbol) have been considered. For each operative
condition and transmission setting, 50 sample records have been acquired and analyzed through the
optimized method. The window function Ollila and an overlap ratio equal to 80% have been adopted.
Examining the obtained results, given in Tab. V and Fig. 5, it can be noted that:
• higher sampling factors do not seem to affect the method’s metrological performance; the same is
true if vertical resolution is considered;
• performance enhancement can be noticed both in the presence of acquired records covering longer
and longer observation periods.
Successively, 50 repeated measurements of total and channel power have been executed by means
respectively of PM and spectrum analyzers (ESA, VSA and RTSA). Tab VI accounts for the results
provided by the PM, while Tab. VII enlists those peculiar to the analyzers. As an example, Fig. 6 sketches
a typical PSD estimated by the proposed method (Fig. 6a), ESA (Fig. 6b), VSA (Fig. 6c) and RTSA
(Fig. 6d).
With regard to total power, the following considerations can be drawn:
• results furnished by the PM are different for the two probes adopted;
• experimental standard deviation peculiar to the PM is slightly better than that assured by the
proposed method;
IM-6241
• PM outcomes concur with total power measurement results of the proposed method; a confidence
level equal to 99% is considered [14].
As for the channel power, it is worth stressing that:
• the proposed method exhibits satisfying repeatability. The related experimental standard deviation is
better than that characterizing ESA, VSA and RTSA results;
• ESA,VSA and RTSA outcomes concur with channel power measurement results of the proposed
method; a confidence level equal to 99% is considered [15]-[17].
V. EXPERIMENTS
A number of experiments on real DVB T signals have been carried out through the optimized method.
The signals have been radiated by two MEDIASET DVB-T multiplexers, operating respectively on
UHF 38 (610 MHz RF central frequency) and UHF 55 (746 MHz RF central frequency) channel.
A simplified measurement station, sketched in Fig. 7, has been adopted. With respect to that used in the
emulation stage, the function generator has been replaced by a suitable amplified antenna, the VSA and
RTSA have been removed and a power splitter has been added. Cables, connectors, and power splitter
have been characterized through the aforementioned vector network analyzer. Mean value and
experimental standard deviation of 100 attenuation measures obtained in the UHF 38 and UHF 55
channels are given in Tab. VIII.
As an example, Fig. 8a and Fig. 8b show the power spectrum of a DVB T signal, radiated by the
MEDIASET multiplexer operating on UHF 55, as estimated by the proposed method and ESA,
IM-6241
respectively. Channel power measurement results are summarized in Tab. IX; good agreement can be
appreciated.
VI. CONCLUSION
Power measurement in DVB-T systems has been dealt with. An optimization stage, properly designed by
the authors, has made a digital signal-processing method, originally addressed to spread spectrum signals,
capable of overcoming low accuracy and repeatability problems currently available instruments suffer
from in the presence of DVB-T signals.
Results of a number of tests conducted on simulated and emulated signals have shown the good
performance of the method. Negligible bias and reduced processing time have been experienced in the
simulation stage, where a reference PSD has been imposed. Experimental standard deviation (as good as
0.2%) comparable to that granted by an RF power meter and better than that characterizing high
performance spectrum analyzers has been observed in the emulation stage. Good agreement between the
results provided by the method and those furnished by a traditional spectrum analyzer in channel power
measurements conducted on real DVB-T signals has confirmed the efficacy of the proposal.
On-going research activity is mainly oriented to the realization of a DVB-T power meter, based on a
digital signal processor and implementing the proposed method.
IM-6241
REFERENCES
[1] “Digital Video Broadcasting (DVB) Framing structure, channel coding and modulation for digital
terrestrial television” ETSI EN 300744, V1.5.1, November 2004
[2] “Digital Video Broadcasting (DVB) Measurement guidelines for DVB systems” ETSI TR 101290,
May 2001.
[3] M.Bertocco, A.Sona “On the power measurement via a spectrum analyzer”, in Proc. of
Instrumentation and Measurement Technology Conference (IMTC 2004), Como, Italy, pp 958-963,
18-20 May 2004.
[4] E.Lawrey, C.J.Kikkert “Peak to average ratio reduction of OFDM signals using peak reduction
carriers”, in Proc. of Fifth International Symposium on Signal Processing and its Application
(ISSPA’99), Brisbane, Australia, pp 737-740, 22-25 August 1999.
[5] M.Bertocco, R.Tittoto, E.Rizzi, L.Benetazzo, “Statistical analysis of measurements for
telecommunication-network troubleshooting”, IEEE Transactions on Instrumentation and
Measurement, vol.52, No.4, pp.1048 1053, August 2003.
[6] L.Angrisani, M.D’Apuzzo, M.D’Arco, “A new method for power measurements in digital wireless
communication systems”, IEEE Transactions on Instrumentation and Measurements, vol. 52, No. 4,
pp. 1097-1106, August 2003.
[7] H.Jokinen, J.Ollila, O.Aumala, “On windowing effects in estimating averaged periodograms of
noisy signals”, Measurement, vol. 28, pp. 197–207, 2000.
[8] P.Moulin, “Wavelet thresholding techniques for power spectrum estimation”, IEEE Transactions on
Signal Processing, vol. 42, pp. 3126–3136, November 1994.
[9] A.T.Walden, D.B.Percival, E McCoy, “Spectrum estimation by wavelet thresholding of multitaper
estimators”, IEEE Transactions on Signal Processing, vol. 46, pp. 3153–3165, December 1998.
IM-6241
[10] D.L.Donoho, I.M.Johnstone, “Ideal spatial adaptation by wavelet shrinkage”, Biometrika, vol. 81,
pp. 425-455, 1994.
[11] D.B.Percival, “Simulating Gaussian random processes with specified spectra”, Comput. Sci. Stat.,
vol. 24, pp. 534–538, 1992.
[12] L.Angrisani, D.Capriglione, L.Ferrigno, G.Miele, “Reliable and repeatable power measurements in
DVB-T systems”, in Proc. of Instrumentation and Measurement Technology Conference (IMTC
2006), Sorrento, Italy, pp 1867-1872, 24-27 April 2006.
[13] “Vector Network Analyzers Technical Data Sheet”, ANRITSU 37100C/37200C/37300C, Rev.C.,
January 2003.
[14] “N1911A and N1912A P-Series Power Meters User’s Guide”, Agilent Technologies P/N N1912-
90002, USA, February 2005.
[15] “Agilent ESA-E Series Spectrum Analyzers Specification Guide”, Agilent Technologies P/N
E4401-90472, USA, April 2004.
[16] “E4406A VSA Series Transmitter Tester User’s Guide”, Agilent Technologies P/N E4406-90177,
USA, September 2001.
[17] “RSA3408A 8 GHz Real-Time Spectrum Analyzer User Manual” (071-1617-01), Tektronix, 2006.
IM-6241
POWER MEASUREMENT
SECTION
DOWNCONVERSION?
DATA ACQUISITION
DOWN CONVERSION
MEASUREMENT ALGORITHMS
WELCH ESTIMATION
MULTITAPER ESTIMATION
PSDESTIMATION
SECTION
INCOMING SIGNAL
N
Y
IF signal
RF signal
PROBE AND ANTENNAS CONDITIONING
SECTION
DIGITIZINGSECTION
RESULTS
DOWNCONVERSION?
DATA ACQUISITION
DOWN CONVERSION
MEASUREMENT ALGORITHMS
WELCH ESTIMATION
MULTITAPER ESTIMATION
POWERMEASUREMENT
SECTION
INCOMING SIGNAL
N
Y
IF signal
RF signal
PROBE AND ANTENNAS CONDITIONING
SECTION
DIGITIZINGSECTION
RF signal
POWER MEASUREMENT
SECTION
DOWNCONVERSION?
DATA ACQUISITION
DOWN CONVERSION
MEASUREMENT ALGORITHMS
WELCH ESTIMATION
MULTITAPER ESTIMATION
PSDESTIMATION
SECTION
INCOMING SIGNAL
N
Y
IF signal
RF signal
PROBE AND ANTENNAS CONDITIONING
SECTION
DIGITIZINGSECTION
RESULTS
DOWNCONVERSION?
DATA ACQUISITION
DOWN CONVERSION
MEASUREMENT ALGORITHMS
WELCH ESTIMATION
MULTITAPER ESTIMATION
POWERMEASUREMENT
SECTION
INCOMING SIGNAL
N
Y
IF signal
RF signal
PROBE AND ANTENNAS CONDITIONING
SECTION
DIGITIZINGSECTION
RF signal
Fig. 1. Block diagram of the adopted measurement method.
IM-6241
0.08
0.09
0.10
0.11
0.12
0.13
0.14
~3 ~6 ~12Oversampling factor
σC [%]6 bits7 bits 8 bits9 bits
0.00000.00100.00200.00300.00400.00500.00600.00700.00800.00900.0100
~3 ~6 ~12Oversampling factor
∆C [%] 6 bits7 bits 8 bits9 bits
0.000
0.100
0.200
0.300
0.400
0.500
0.600
0.700
~3 ~6 ~12Oversampling factor
∆T [%] 6 bits7 bits 8 bits9 bits
a) b)
c) d)
Fig. 2. Simulation stage: a) σC, b) ∆C, and c) ∆T versus ENOB for three values of the oversampling factor; d) estimated PSD for the considered values of ENOB.
IM-6241
0.14
0.19
0.24
0.29
0.34
1/4 1/2 3/4 1Acquired record length
σT [%]∼3
∼6
∼12
a)
b)
0.00000.00500.01000.01500.02000.02500.03000.0350
1/4 1/2 3/4 1Acquired record length
∆T [%]~3~6~12
Fig. 3. Simulation stage: a) σT and b) ∆T versus acquired record length for three values of the oversampling factor.
IM-6241
Fig. 4. Measurement station for performance assessment.
IM-6241
0,043 ~3
Fig. 5. Emulation stage: σT versus acquired record length for three values of the oversampling factor.
0,008
0,013
0,018
0,023
0,028
0,033
0,038
1/4 1/2 3/4 1Acquired record length
σT [ µ W]~8~16
IM-6241
22:27:38 Apr 6, 2006
Center Freq 610.000000 MHz
Start Freq 605.000000 MHz
Stop Freq 615.000000 MHz
CF Step 1.00000000 MHzAuto Man
Freq Offset 0.00000000 Hz
Signal TrackOn Off
Scale TypeLog Lin
Freq/Channel
Ref -25 dBm Atten 5 dBPeakLog8dB/
W1 S2S3 FC
AA
Center 610 MHzRes BW 100 kHz VBW 100 kHz
Span 10 MHzSweep 4 ms (401 pts)
Center610.0000000 MHz
0
a) b)
Print ToFile Printer
File Typewmf
File LocationA: C:
ImageInvert Normal
HCOPy DestPrinter
Print Setup
Center Freq 610.000000 MHz
R L T S
Ref Lvl-45.00 dBm 8.00
dB/
Center 610.000 MHz Span 10.0000 MHzRes BW 871.0 Hz
Trig Free
ExtAt 0.0
MaxP -15.0
Channel Power Power Spectral Density
-20.72 dBm/ 8.00000 MHz -89.75 dBm/Hz
03/31/70 23:57:32 Basic
Channel PowerBase Ch Freq 610.000 MHz
Fig. 6. Power spectrum of an emulated DVB-T signal estimated by a) the proposed method, b) ESA, c) VSA and d) RTSA.
605 606 607 608 609 610 611 612 613 614 615
-100
-80
-60
-40
power spectrum [dBm/Hz]-20
frequency [MHz]
c) d)
IM-6241
Fig. 7. Measurement station for the experiments on real DVB-T signals.
IM-6241
741 742 743 744 745 746 747 748 749 750 751-135
-125
-115
-105
-95
-85
-75
-65
-55
frequency [MHz]
pow
er s
pect
rum
[dB
m/H
z]
a)
b)
Fig. 8. Power spectrum of a real DVB-T signal measured by the a) proposed method and b) ESA.
IM-6241
Tab. I. DVB-T measurement parameters and their applicability according to the ETSI TR 101.290 (T=transmitter, N=network, R=receiver).
Measurement parameter T N R RF frequency accuracy (precision) X Selectivity X AFC capture range X Phase noise of local oscillators X X RF/IF signal power X X Noise power X RF and IF spectrum X Receiver sensitivity/ dynamic range for a Gaussian channel
X
Equivalent Noise Degradation (END) X Linearity characterization (shoulder attenuation) X Power efficiency X Coherent interferer X BER vs. C/N ratio by variation of transmitter power X X BER vs. C/N ratio by variation of Gaussian noise power
X X
BER before Viterbi (inner) decoder X BER before RS (outer) decoder X BER after RS (outer) decoder X I/Q analysis X X Overall signal delay X SFN synchronization X Channel characteristics X
IM-6241
Tab. II. Results obtained in the simulation stage: multitaper estimator is involved.
Oversampling factor
Figure of merit Guard interval [µs] ~3 ~6 ~12
28 0,148 (sym8,10)
0,176 (db3,25)
0,151 (db3,25)
σT [%]
224 0,129 (coif1,50)
0,097 (coif1,50)
0,117 (db3,25)
28 0,148 (sym8,10)
0,176 (db3,25)
0,151 (db3,25)
σC [%]
224 0,129 (coif1,50)
0,097 (coif1,50)
0,117 (db3,25)
28 0,1104 (bior2.8,50)
0,1252 (bior2.8,50)
0,6335 (bior2.8,50)
∆T [%]
224 0,1509 (bior2.8,50)
0,1050 (bior2.8,50)
0,1237 (bior2.8,50)
28 0,1105 (bior2.8,50)
0,1253 (bior2.8,50)
0,6336 (bior2.8,50)
∆C [%]
224 0,1509 (bior2.8,50)
0,1050 (bior2.8,50)
0,1238 (bior2.8,50)
28 12,55 59,46 280,53 Measurement time [s]
224 53,61 280,58 1168,20
IM-6241
Tab. III. Results obtained in the simulation stage: WOSA estimator is involved.
Oversampling factor
Figure of merit Guard interval [µs] ~3 ~6 ~12
28 0,149 (Ollila,70)
0.181 (Ollila,50)
0,148 (Ollila,30)
σT [%]
224 0,130 (blackman,60)
0,098 (hanning,50)
0,092 (Ollila,50)
28 0,149 (Ollila,70)
0,181 (Ollila,50)
0,148 (Ollila,30)
σC [%]
224 0,130 (blackman,60)
0,098 (hanning,50)
0,092 (Ollila,50)
28 0,0015 (FD3FT,30)
5,4091e-4 (FD3FT,30)
0,0017 (FD4FT,60)
∆T [%]
224 0,0068 (blackman,40)
0,0028 (MS3FT,10)
0,0020 (MS4FT,60)
28 0,0015 (FD3FT,30)
6,0611e-4 (FD3FT,30)
0,0012 (FD4FT,60)
∆C [%]
224 0,0068 (blackman,40)
0,0017 (MS3FT,10)
0,0020 (MS4FT,60)
28 0,032 0,052 0,096 Measurement time [s] 224 0,053 0,094 0,177
IM-6241
Tab. IV. Characterization results of cables and connectors utilized in the measurement station of Fig.4.
Mean
attenuation
Experimental standard deviation
Power meter 0,829150 0,000039 Spectrum analyzers 0,834860 0,000019
Oscilloscope 0,834140 0,000014
IM-6241
Tab. V. Total and channel power measures provided by the proposed method. The acquired record covers a single DVB-T symbol.
8k transmission mode, 64-QAM modulation scheme, 8 bit vertical resolution
Oversampling factor
Figure of Merit Guard Interval [µs] ~3 ~8 ~16
28 0.012 0.014 0.013 σT [µW]
224 0.0094 0.017 0.011
28 0.012 0.014 0.013 σC [µW]
224 0.0094 0.017 0.011
28 9.931 10.024 9.937 PT [µW]
224 10.142 10.163 10.144
28 9.890 9.989 9.895 PC [µW]
224 10.1030 10.125 10.105
8k transmission mode, 64-QAM modulation scheme, 7 bit vertical resolution
Oversampling factor
Figure of Merit Guard Interval [µs] ~3 ~8 ~16
28 0.011 0.034 0.014 σT [µW] 224 0.011 0.029 0.015 28 0.011 0.032 0.014 σC [µW] 224 0.011 0.028 0.016 28 10.148 10.162 10.157 PT [µW] 224 10.079 10.098 10.097 28 9.971 9.985 9.980
PC [µW] 224 9.899 9.919 9.916
IM-6241
Tab. VI. Mean values (PPM) and experimental standard deviations (σPM) of total power measures provided by the PM equipped with N1921A and E9304A probes.
Transmission Settings 8k, 64-QAM, 610 MHz central frequency
PM Guard
Interval [µs]
PPM
[µW] σPM
[µW]
28 9.9444 0.0018 N1921A PROBE 224 9.9630 0.0020
28 8.0402 0.0060 E9304A PROBE 224 7.95910 0.00086
IM-6241
Tab. VII. Mean values (PSA) and experimental standard deviations (σSA) of channel power measures provided by ESA, VSA and RTSA; different settings of their resolution bandwidth have been considered.
Transmission Settings: 8k, 64-QAM, 610 MHz central frequency
Instrument RBW [kHz]
Guard Interval [µs]
PSA [µW]
σSA [µW]
100 28 10.322 0.074
100 224 10.656 0.080
30 28 10.376 0.068 ESA
30 224 10.142 0.070
0.871 28 10.506 0.036
0.871 224 10.218 0.023
30 28 10.162 0.099 VSA
30 224 9.52 0.12
50 28 9.311 0.044
50 224 9.318 0.042
30 28 9.158 0.041 SPECTRUM
ANALYZERS
30 224 9.042 0.044
28 9.177 0.097
RTSA
REAL TIME MODE
224 9.088 0.081
IM-6241
Tab. VIII. Characterization results of cables and connectors utilized in the measurement station of Fig. 7.
UHF Channel
Mean attenuation
[dB]
Experimental standard deviation
[dB] 38 -4.703 0.032 DAS 55 -5.403 0.042 38 -19.393 0.021 Traditional Spectrum
analyzer 55 -19.3886 0.0086
IM-6241
Tab. IX. Experimental results.
8k transmission mode, 64-QAM, 28µs guard interval UHF Channel 38
610 MHz UHF Channel 55
746 MHz Proposed Method 90.94 nW 93.07 nW
Traditional Spectrum Analyzer 94.06 nW 93.23 nW
IM-6241
List of figure and table captions
Fig. 1. Block diagram of the adopted measurement method.
Fig. 2. Simulation stage: a) σC, b) ∆C, and c) ∆T versus ENOB for three values of the oversampling factor; d) estimated PSD for the considered values of ENOB.
Fig. 3. Simulation stage: a) σT and b) ∆T versus acquired record length for three values of the oversampling factor.
Fig. 4. Measurement station for performance assessment.
Fig. 5. Emulation stage: σT versus acquired record length for three values of the oversampling factor.
Fig. 6. Power spectrum of an emulated DVB-T signal estimated by a) the proposed method, b) ESA, c) VSA and d) RTSA.
Fig. 7. Measurement station for the experiments on real DVB-T signals.
Fig. 8. Power spectrum of a real DVB-T signal measured by the a) proposed method and b) ESA.
Tab. I. DVB-T measurement parameters and their applicability according to the ETSI TR 101.290 (T=transmitter, N=network, R=receiver).
Tab. II. Results obtained in the simulation stage: multitaper estimator is involved.
Tab. III. Results obtained in the simulation stage: WOSA estimator is involved.
Tab. IV. Characterization results of cables and connectors utilized in the measurement station of Fig.4.
Tab. V. Total and Channel power measures provided by the proposed method. The acquired record covers a single DVB-T symbol.
Tab. VI. Mean values (PPM) and experimental standard deviations (σPM) of total power measures provided by the PM equipped with N1921A and
E9304A probes.
Tab. VII. Mean values (PSA) and experimental standard deviations (σSA) of channel power measures provided by ESA, VSA and RTSA; different
settings of their resolution bandwidth have been considered.
Tab. VIII. Characterization results of cables and connectors utilized in the measurement station of Fig. 7
Tab. IX. Experimental results.