DMR and P25 digital radio standards are specifically designed
to provide voice coverage very similar to their analog
narrowband FM predecessors. This means that migration
from analog narrowband FM voice systems to more spectrally-efficient, data-capable DMR or P25 networks does not require
additional repeater sites.
However, when specifying or designing digital radio networks for
data transmission, there are important coverage differences that
you need to understand, to be sure you specify for the overall
network performance you need.
First, let’s define some typical performance metrics:
● Voice network performance is usually defined as a Delivered
Audio Quality (DAQ);
● data performance is normally defined by Message Error
For voice, a DAQ of 3. 4 (speech understandable without
repetition, although some noise or distortion present) is
acceptable for most Public Safety applications. This equates
to a Bit Error Rate (BER) of around 2% – two out of every 100
bits received are decoded incorrectly. The decoded audio is
noticeably distorted, but is still perfectly understandable to the
Things are not so simple with data, unfortunately. For example,
if a data message is 1000 bits long, then that same 2% BER will
correspond to an average of 20 errors. Even with Forward Error
Correction (FEC) techniques, a BER of 2% means few (if any)
1000-bit messages will be received correctly. That 2% BER
equates to nearly 100% MER!
Let’s look at it the other way round. If, for example, we decide
that MER of 1% is acceptable, that means only one 1000-bit
message in 100 is received incorrectly – conceivably, only one
bit in 100,000 has been received incorrectly. So our 1% MER
equates to a BER of just 0.001%.
In terms of the acceptable received signal level, for a DMR
voice network to achieve DAQ3.4, the signal must be at least
15.6dB above the noise and interference present, the carrier-to-noise ratio (CNR) must be greater than 15.6dB. For a DMR data
network, the minimum CNR will depend on message length; the
longer the message, the greater the CNR required to achieve the
desired 1% MER.
This reinforces an important difference: for DMR voice there is
a single CNR threshold, whereas DMR data networks must be
designed with the longest message length in mind, because the
threshold varies with message length. Typically, the acceptable
CNR for DMR data performance needs to be 10–15dB higher than
it does for DMR voice.
Now let’s bring it back to what really matters—coverage. If we
look at line-of-sight coverage according to free space path loss
(FSPL), and the data network requires CNR 10dB higher than
voice to achieve 1% MER, the resulting coverage radius will only
be 30% of that for DAQ3.4 DMR voice (or 10% of the covered
area). For longer message lengths – where the CNR requirement
is 15dB higher than for DMR voice – the coverage radius will only
be 17% of that for voice. That’s just 3% of the covered area.
WHAT DOES THIS MEAN IN REAL TERMS?
Luckily, the news isn’t quite as bad as these theoretical figures
suggest. In practice, over real terrain with no line of sight,
the covered area differences would be less, but would still
be significant and will vary depending on the type of terrain.
Similarly, there are technical solutions that can narrow the
coverage gap further, e.g. involving technology selection system
design, and at product level. That’s why it’s critical that you
consult coverage experts when you are investing in a new
digital radio network.
2000 errors per 100,000 bits = 2% BER
1000 2000 3000 4000 99,000 100,000
1 error per 100,000 bits = 0.001% BER
1000 2000 3000 4000 99,000 100,000