Public Information
Dokumentnamn - Document name
Sida - Page
1(4)
OVERVIEW
Uppgjord - Prepared
Tfn - Phone
LU/EPL/T/RP Mats Nordlund
Godknd - Approved
0920 - 20 20 78
Kontr - Checked
Datum - Date
1999-09-05
Rev
Dokumentnr - Document no
EPL/T/R-99:150
Tillhr/referens - File/reference
LU/EPL/T/RP Gunnar Heikkil
Measuring C/I in TEMS Investigation
Abstract
With the new release of TEMS Investigation, the users are provided with
a new measure of the radio network performance, namely the carrier to
interference ratio of the received radio signal, also referred to as C/I.
This makes it possible to identify frequencies exposed to high levels of
radio interference
Background and motivation
Until now, operators have had a means of supervising their radio networks downlink quality by using the speech quality index (SQI) present
in TEMS. Measurements of SQI help identify geographical areas with
inadequate speech quality. However, if frequency hopping is applied in
such areas, it has traditionally been difficult for the users to determine
which of the frequencies are disturbed and why the speech quality is
unsatisfactory. To help resolve such ambiguities, TEMS users are now
provided a means of measuring the average C/I for each of the frequencies used in an on-going call.
In the past, rough C/I measurements have been possible to perform in
some cases by comparing signal power measurements for the serving
cell and neighbouring cells which use the same traffic channels as used
in the serving cell (but different BCCHs). This type of measurement does
only compare the BCCHs signal powers and does not regard whether
the TCHs might use power control and/or DTX.
Utilization of prediction tools such as TEMS Cellplanner results in frequency plans that are supposed to be optimized in terms of C/I. These
tools use detailed descriptions of the topology in combination with
advanced radio propagation models to produce the predicted C/I values. However, features like DTX and power control yield high error margins for the predictions. As a consequence, areas with low C/I levels
might be missed.
TEMS Investigation makes it possible to obtain measurements of the
actual C/I as experienced by the mobile station. Furthermore, TEMS
Investigation performs measurements of C/I for each of the frequencies
utilized for communication.
Public information
Dokumentnamn - Document name
Sida - Page
2(4)
OVERVIEW
Uppgjord - Prepared
Tfn - Phone
LU/EPL/T/RP Mats Nordlund
Datum - Date
Rev
0920 - 20 20 78 1999-09-05
Dokumentnr - Document no
EPL/T/R-99:150
Measurement method
With continuous transmission from the communicating base station, the
TDMA radio bursts enter the mobile stations air interface at a rate of 100
bursts per half second. The average C/I is obtained by applying
advanced signal processing algorithms to each of the received radio
bursts.
Relating to the protocol stack, the measurement is performed within
layer 1 - the physical layer. As shown in figure 1, the carrier and interferer signals are summed before entering the receivers antenna. After
analog-to-digital conversion, the composite signal is represented in the
digital domain. The signal processing algorithms are applied to the digital signal in order to measure the ratio between the power of the desired
signal (the carrier) and the power of interfering signal.
digital domain
synchronization
C
RF receiver
& A/D
error correction
C/I
Figure 1.
data demodulation
to speech codec, etc.
Simplified illustration of where the measurement of C/I takes place.
In dedicated mode, TEMS Investigation presents the average C/I twice
a second, which is equal to the ordinary measurement interval. If frequency hopping is employed, the average C/I for each frequency is presented.
It should be noted that the measurement method does not regard what
I actually consists of. Normally, in interference limited scenarios, I
reflects the co-channel interference. However, in coverage limited (noise
limited) scenarios, C/I rather reflects the carrier to noise ratio, C/N.
Drive test example
To illustrate the usage of C/I measurements, the results from a test drive
in a live network is shown in figure 2. The test drive lasts for 40 seconds
and utilizes cyclic frequency hopping over four frequencies. Enhanced
full rate speech coding is used throughout the test drive. The top graph
shows TEMS SQI and RxLev. The bottom graph shows the C/I measurements for the four frequencies used. As can be seen in figure 2, the SQI
Public information
Dokumentnamn - Document name
Sida - Page
3(4)
OVERVIEW
Uppgjord - Prepared
Tfn - Phone
LU/EPL/T/RP Mats Nordlund
Datum - Date
0920 - 20 20 78 1999-09-05
Rev
Dokumentnr - Document no
EPL/T/R-99:150
curve dips sharply at the end of the test drive. RxLev alone does indicate
that the signal power level is about 50 dB above -110 dBm. Hence, it is
concluded that the dip in quality does not depend on low signal power
level, that is, the quality problem is related interference rather than coverage. An interesting observation is that RxLev in fact increases during
the SQI dip. The reason is most likely due to increasing interfering
power.
20
60
SQI
50
0
0
RxLev
40
10
15
20
25
time [s]
30
35
40
10
15
20
25
time [s]
30
35
40
RxLev [dB]
SQI [dBQ]
By looking at the C/I curves it is seen that two of the four frequencies
(shown bold) are exposed to C/I levels around 5 dB during the SQI dip.
As can be seen in figure 2, with two out of the four frequencies below 10
dB, the resulting speech quality is heavily deteriorated.
C/I [dB]
25
20
15
10
5
0
0
Figure 2.
Live measurements of SQI, RxLev and C/I
By taking corresponding positioning data into account, this information
may then be regarded when optimizing the frequency plan of the specific area.
Measurement performance
The C/I measurement range is from -5 dB to +25 dB. Outside this range
the measurement method saturates. Average values lower than -5 dB are
considered highly unlikely to encounter. In addition, when the number
of hopping frequencies is low, average C/I levels below the lower limit
normally result in a dropped call. Above the higher limit, the perceived
speech quality is not further improved. Hence, the limited measurement
range is not a restriction.
Public information
Dokumentnamn - Document name
Sida - Page
4(4)
OVERVIEW
Uppgjord - Prepared
Tfn - Phone
LU/EPL/T/RP Mats Nordlund
Datum - Date
0920 - 20 20 78 1999-09-05
Rev
Dokumentnr - Document no
EPL/T/R-99:150
If downlink DTX is utilized, the number of transmitted bursts (from the
base station to the mobile station) may be lower than the maximum 100
depending on the speech activity of the transmitting side. TEMS Investigation performs measurements on those bursts actually sent from the
base station and disregards non-transmitted bursts.
The number of hopping frequencies determines the number of bursts
used for each frequencys C/I measurement. For example, if four frequencies are used, on average 25 bursts per frequency are received each
0.5s interval. The number of samples per frequency each 0.5s interval
decreases with increasing number of hopping frequencies. The implication is that the measurement accuracy is better for small number of hopping frequencies compared to high number of frequencies.
If the true C/I is within the range 0 dB to 15 dB, and four frequencies are
used for transmission (without DTX interruptions all bursts usable),
the measurement error is typically better than 1 dB.
