Page 263 - Electrical Engineering Dictionary

power spectral density as well as all finite- 2. If X is a gray-level image on a space E and
order joint moments. B is a subset of E, for every p ∈ E we have

ergodicity stochastic processes for which (X B)(p) = inf X(p + b)
b∈B
ensemble averages can be replaced by tem-
poral averages over a single realization are 3. If both X and B are gray-level images on
said to be ergodic. For a stochastic pro- a space E, for every p ∈ E we have
cess to be ergodic, a single realization must

in the course of time take on configurations (X B)(p) = inf X(p + h) − B(h)
h∈E
closely resembling the entire ensemble of
processes. Stationary filtered white noise is with the convention ∞ − ∞=+∞ when
considered ergodic, while the sinusoidal pro- X(p + h), B(h) =±∞. (In the two items
cess A cos(wt +φ) with random variables A above, X(q) designates the gray-level of the
and φ is not. point q ∈ E in the gray-level image X.) See
dilation, structuring element.
Erlang B a formula (or mathematical
model) used to calculate call blocking prob- ERP See effective radiated power.
ability in a telephone network and in partic-
ular in cellular networks. This formula was error (1) manifestation of a fault at log-
initially derived by A. K. Erlang in 1917, a ical level. For example, a physical short or
Danish pioneer of the mathematical model- break may result in logical error of stuck-at-0
ing of telephone traffic, and is based on the or stuck-at-1 state of some signal in the con-
assumption that blocked calls are forever lost sidered circuit.
to the network. See also Erlang C. (2) a discrepancy between a computed,
observed, or measured value or condition and
Erlang C similar to Erlang B, the for- the true, specified, or theoretically correct
mula is based on a traffic model where the value or condition. See bug,
call arrival process is modeled as a Poisson exception.
process, the call duration is of variable length
and modeled as having an exponential dis- error control coding See channel coding.
tribution. The system is assumed to have a
queue with infinite size that buffers arriving
calls when all the channels in the switch are error-correcting code (ECC) code used
occupied. The model is based on the assump- whencommunicationdatainformationinand
tionthatblockedcallsareplacedinthequeue. between computer systems to ensure correct
data transfer. An error correcting code has
Erlang capacity maximum number of enough redundancy (i.e., extra information
users in the system which leads to the maxi- bits) in it to allow for the reconstruction of
mum allowable blocking probability (for ex- the original data, after some of its bits have
ample, 2%). been the subject of error in the transmission.
The number of erroneous bits that can be re-
erosion an important basic operation in constructed by the receiver using this code
mathematical morphology. Given a structur- depends on the Hamming distance between
ing element B, the erosion by B is the oper- the transmitted codewords. See also error
ator transforming X into the Minkowski dif- detecting code.
ference X B, which is defined as follows:
1. If both X and B are subsets of a space E, error correction capability of a code is
bounded by the minimum distance and for an
X B ={z ∈ E |∀b ∈ B, z + b ∈ X} (n, k) block code, it is given by t =[(d min −

c
2000 by CRC Press LLC