Discrete Estimations of Cross-Correlation Components
of Periodically Correlated Random Signals
I. N. Yavorskyj
1,2,*
, R. Yuzefovych
1,**
, I. Y. Matsko
1
, and Z. Zakrzewski
2
1
Karpenko Physico-Mechanical Institute of NASU, Lviv, Ukraine
2
University of Technology and Life Sciences, Bydgoszcz, Poland
*e-mail: iavor@ipm.lviv.ua
**e-mail: abzac@ipm.lviv.ua
Received in final form December 9, 2013
Abstract—Properties of cross-correlation components estimations of two periodically correlated
random signals are analyzed. The conditions of absence of the first and the second genus aliasing effects
are obtained. The formulas for estimations variance which allow to select a valid sampling step in
dependences on realization length and signal properties are derived. The results are concretized for
amplitude modulated signals.
DOI: 10.3103/S0735272714020034
Sampling is necessary procedure for statistical signals processing by means of computer equipment. One
of the most important problem for its realization is a selection of sampling step. Traditionally this step is
selected using Shannon–Kotelnikov theorem and its generalizations. Correspondent the least sampling
frequency is called Nyquist frequency.
In this connection there is a question, whether the sampling step, selected in such way, satisfies the
conditions for solution the problems of signal statistical analysis. Especially it is concern of statistics of
non-stationery processes, where it is necessary to operate the correlation functions of two variables even in
the second order theory, and these functions behavior properties can be different with regard to each
variable.
In this paper there are represented results of analysis of discrete estimations of Fourier coefficients of
cross-correlation function of periodically correlated random processes (PCRP), i.e. mathematical model of
the signals, describing their stochasticity, and repeatability of their properties [1–5]. Properties of
repeatability and stochasticity are typical for most signals applying in the systems of communication,
telemetry, radio and hydrolocation. These properties are achieved by signals after processes of modulation
scanning, encoding, antenna rotation, etc. [3–5].
Mathematical expectation of PCRP is m t E t
x
x () () = and its correlation b tu E t t u
x
x x (,) ()( ) = +
o o
,
x x
x
o
() () () t t m t = - are time periodical functions m t m t T
x x
() ( ) = + , b tu b t Tu
x x
(,) ( ,) = + . Signals models in
form of stationery random processes can be considered as stationery approximation of PCRP: their
characteristics are determined by time averaging of the last characteristics. Application of periodical
non-stationerity gives an opportunity for creation of such facilities for detection, filtering, classification, and
also estimation the parameters, exceeding analogous ones, based on stationery approach [3, 4].
Analysis of single- and multi-channel communication systems, identification random processes
propagation paths, definition of their destinations and sources require analysis of interrelation of two or
several PCRP. In [6] there are represented main properties of cross- correlation function of such signals, and
also properties of coherent estimations by continuous realization was carried out. This paper develops these
results on discrete case.
Mutual correlation function of interrelated PCRP b tu E t t u
xh
x h (,) ()( ) = +
o o
, h h
h
o
() () () t t m t = - ,
m t E t
h
h () () = is periodical time t function b t Tu b tu
xh xh
( ,) (,) + = and it can be represented in form of
Fourier series:
78
ISSN 0735-2727, Radioelectronics and Communications Systems, 2014, Vol. 57, No. 2, pp. 78–91. © Allerton Press, Inc., 2014.
Original Russian Text © I.N. Yavorskyj, R. Yuzefovych, I.Y. Matsko, Z. Zakrzewski, 2014, published in Izv. Vyssh. Uchebn. Zaved., Radioelektron., 2014, Vol. 57,
No. 2, pp. 29–42.