Professional Documents
Culture Documents
April 2011
If someone asks you "Which Signal Level for good call quality: -80 dbm or -90 dBm?"
Beware, if you respond quickly, you might end up missing. This is because the correct
answer is ... it depends! The Signal Strength is a very important and essential measure for
any technology (GSM, CDMA, UMTS, LTE, etc.). However, it is not the only one: let's talk a
little today about another magnitude, equally important: the Signal Noise Ratio.
Although this ratio is of fundamental importance to any cellular system, is not well
understood by many professionals. On the opposite side, professionals with a good
understanding of this ratio are able for example, to correctly assess the RF links, and also to
perform more extensive optimizations, obtaining the best possible performance of the
system.
Eb and No
To begin, we define the basic concepts of Eb and No. They are basic for any digital
communication system, and generally we talk about it when we deal with Bit Error Rate and
also Modulation techniques.
Simply put:
Unit: dB
Do not worry. Indeed, only with the theoretical definition is still very difficult to see how this
ratio
is
used,
or
how
it
can
be
measured.
But okay, let's walk a little further.
The ratio Eb/No is measured at the receiver, and serves to indicate how strong the signal is.
Depending on the modulation technique used (BPSK, QPSK, etc.) we have different curves
for Bit Error Rate x Eb/No.
These curves are used as follows: for a certain RF signal, which is the bit errors rate that I
have? Is this bit error rate acceptable for my system?
Whereas the gain that digital has, then we can set a minimum criterion of signal to noise
ratio, in order to have each service (Voice/Data) operating acceptably.
In other words, we can theoretically determine how the performance would be for the digital
link.
Note: it is worth remembering here that this is a very complex subject. As always, we try to
introduce to you the most simplified possible through the use of examples and simple
concepts. Okay?
For example, a concept that could be explored here - since we are talking about digital
communication system - is the Noise Figure. But we do not want to repeat here all the
theory explained in the University. Nor was it to have mentioned the noise figure here, but
as we talked about it, just understand as a noise level that every receiver has, and that it is
due to the process of amplification and processing of signal.
Concepts like this, and other even more complex, can be studied, if you wish. But now, let's
continue with our signal to noise ratio.
To proceed, we must know some more definitions. However, this point is quite delicate, as
we enter a conceptual area where we have differences between authors, differences in
translations/countries, where differences in technologies are applied, etc..
E: is the signal (average) energy - do not confuse it with the sinal (average)
power.
b, c, s. ..: Energy are the power points in time, therefore related to the
measure or 'length' of the time (the average power is independent of time ).
o
Hence it comes Eb, Ec and Es, respectively relating to Bit Chip and
Symbol in different times.
Note: With these concepts, several formulas can be derived with different numerators and
denominators. For example, Es = Eb * k, where k = number of bits per symbol. In QPSK
modulation, where k = 2, Es = 2 * Eb. And the derivations of formulas can reach far more
complex equations, such as the definitions of capacity of an AWGN channel, and further
deductions for equivalences (Ec/No, Eb/Nt, etc. ...). Again, it is not our purpose here today.
We only mention a few concepts, related.
Then come back to the practical level - noting that theoretical approaches can be done more
easily later, after the basics are understood.
So let's keep today in ratios most common: Eb/No and Ec/Io.
As we defined Eb/No is the Average Energy of a bit signal, on the Spectral Density of Noise.
It is primarily a parameter related to the manufacturer for different bearers (based on the
channel model). But it can also vary with the environment (urban, rural, suburban), speed,
diversity, use of power control, application type, etc..
And now we can begin to define Ec/Io, one of the most important systems in CDMA and
UMTS.
Note: An important observation is that often when we refer to Ec/Io, we are actually
referring to Ec/(Io + No). What happens is that for practical purposes, we only have Ec/Io,
because the interference is much stronger and the noise can be neglected. Otherwise: for
CDMA interference is like a noise, then both can be considered the same thing.
Okay, let's stop with the issues and concepts, and talk a little about the values of these
indicators and their use in practice.
In terms of values, and talking logarithmicly, if any ratio is less than 1, then the value is
negative. If greater than 1, positive.
We have Ec/Io in the air, which is spread across the spectrum: then we have negative value
to the ratio of energy on the total noise (the energy is lower than the Total Interference). It
is measured at the input of receiver (NodeB, UE, etc).
Regarding Eb/No, it is in the baseband after despreading and decoded only for one user then we have a positive amount of energy over the total noise. It is measured at the output
of receiver (NodeB, UE, etc).
Thus we have:
Ec = 2 W
Io = 0 + 2 + 2 = 4 W
Ec/Io = (2/4) = 0.5 = -3 dB
Now assume that several traffic channels are busy (eg use 6 W for traffic channels). This is
a situation of traffic load, we'll see how is Ec/Io.
Ec = 2 W
Io = 2 + 2 + 6 = 10 W
Ec/Io = (2/10) = 0.2 = -7 dB
Conclusion: As the traffic load in the sector increases, the Ec/Io worsens.
Let us now consider another situation. Instead of one, we have five sectors signal arriving at
the mobile (for simplicity, all with the same level of -90 dBm).
Now have Io = -83 dBm (which is the sum of five signals of -90 dBm). And the power of our
pilot channel remains the same (Ec = -96 dBm).
Thus: Ec/Io = -96 - (-83) = -13 dB
Conclusion: As many more sectors serves the mobile, the Ec/Io worsens.
This situation where we have many overlapping sectors, and with the same level of signal is
known as Pilot Pollution - the mobile sees them all at once - each acting as interferer to
each other.
The solution in such cases is to eliminate unwanted signals, by setting power parameters or
physical adjustments (tilt, azimuth), leaving just dominant signals which should exist at this
problematic place.
The value of Ec/Io fluctuates (varies), as well as any wireless signal. If the value starts to
get too low, you start to have dropped calls, or can not connect. But what then is a good
range of Ec/Io for a sign?
In practical terms, values of Ec/Io for a good evaluation of the network (in terms of this
indicator) are shown in the diagram below.
Conclusion
Today we had a brief introduction on the Ec/Io ratio, a measurement for decisions in CDMA
and UMTS, and used togheter with the measured Signal Strength.
We have seen that it represents the ratio of signal energy within the duration of a chip of
the pilot channel, on the Spectral Density of Noise + Interference.
This is a very important measure, which somehow ignores the overall strength of the signal,
and focuses on how best to evaluate the pilot channel signal is desired, in relation to noise
that interferes with it.
Returning to our original question: A strong signal level does not necessarily indicate an
strong Ec/Io: it depends on the level of interference.