Professional Documents
Culture Documents
\
|
=
D
ECE Department Florida Institute of Technology
Page 36
Signal Processing
From Voice to Radio Waves
Sampling,
Quantization and
source encoding
Channel
Encoding
(Error Correction
Coding)
Interleaving
Burst
Formating
Mapping
De-
Ciphering
Modulation
De-Modulation
Ciphering
Burst
Formating
Mapping
De
-Interleaving
Channel
Decoding
(Error Correction )
Source Decoding
and Wavef orm
Generation
Um
Interf ace
Voice
Signal
Voice
Signal
Transmit Side
Receive Side
As a digital TDMA technology GSM implements extensive
signal processing
ECE Department Florida Institute of Technology
Page 37
Sampling and Quantization
Sampling
o Sampling theorem
specifies conditions for
discretization of band
limited analog signals
o Voice needs to be
sampled at the sampling
rate greater then
8000Hz
Quantization
o Discrete values
assigned to continuous
samples
o Quantization noise
o In GSM, voice is
sampled at 8 K
samples/sec and
quantized with 8192
levels (13 bit words)
111 +3V
110 +2V
101 +1V
0V
001 -1V
010 -2V
011 -3V
111 +3V
110 +2V
101 +1V
0V
001 -1V
010 -2V
011 -3V
Analog Signal
Sampling Pulse
PAM
101 110 101 100 010 010 010 100 111 111
PCM
ECE Department Florida Institute of Technology
Page 38
Speech Source Encoding
Speech coder reduces the data rate
needed for voice signal representation
GSM specifies operation of :
o Full rate vocoder
13Kb/sec
o Half rate vocoder
5.6Kb/sec
o Enhanced Full Rate (EFR)
12.2Kb/sec
o AMR (Adaptive multi rate)
AMR-FR (4.75-12.2Kb/sec)
AMR-HR (4.75-7.95Kb/sec)
AMR rate - function of C/I
BPF
A/D
converter
SPEECH
ENCODER
CHANNEL
CODING
TO
MODULATOR
MICROPHONE
BAND-PASS
300 Hz-3.4 kHz
SPEECH
DECODER
CHANNEL
DECODER
LP
LOW-PASS
4 kHz
D/A
converter
Vocoders enable efficient channel
utilization
ECE Department Florida Institute of Technology
Page 39
Performance comparison of some
commercial vocoders
Mean Opinion Scores (MOS) - Voice Quality
source IIR. The First Annual CDMA Congress
London, Oct. 29-30, 1997
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Clean Speech 20dB SNR
Babble
20dB SNR
Car
15dB SNR
Street
Mu-PCM
8Kb/s EVRC
(CDMA)
13Kb/s CELP
(CDMA)
IS-136 ACELP
GSM EFR
ECE Department Florida Institute of Technology
Page 40
Error control coding (ECC) increases the robustness of the
signal
ECC increases the overhead and reduces the efficiency of the
communication
In GSM, the ECC increases the overhead per user by 57%
Channel Encoding
TYPE Ia
BITS
TYPE II
BITS
TYPE Ib
BITS
CONVOLUTIONAL
ENCODER
r=1/2
K=5
M
U
X
ERROR DETECTING CODE
50
132
78
3
4
189
189
378
456
0
TO
INTERLEAVER
F
R
O
M
V
O
C
O
D
E
R
ECE Department Florida Institute of Technology
Page 41
Interleaving
In mobile
communications, the
errors are bursty
Optimal performance
from ECC is obtained for
uniform error
distribution
Interleaving increases
the performance of ECC
in mobile environment
(
(
(
(
(
(
25 20 15 10 5
24 19 14 9 4
23 18 13 8 3
22 17 12 7 2
21 16 11 6 1
b b b b b
b b b b b
b b b b b
b b b b b
b b b b b
Data is written
column-wise
Data is read
row-wise
Interleaver
b
1
b
2
b
3
b
4
b
5
b
6
b
7
b
8
b
9
b
10
b
11
b
12
b
13
b
14
b
25
b
16
b
17
b
18
b
19
b
20
...
b
1
b
6
b
11
b
16
b
21
b
2
b
7
b
12
b
17
b
22
b
3
b
8
b
13
b
18
b
23
b
4
b
9
b
14
b
19
b
24
..
Burst Error
Caused by
Rayleigh Fading
Errors are spread over the bit stream
ECE Department Florida Institute of Technology
Page 42
Modulation: GMSK (Gaussian MSK)
GMSK has excellent spectral
characteristics
o Low sidelobes
o Robust to non- linearities
Price paid is in the increased
Inter Symbol Interference (ISI)
Simplified GMSK block diagram
MSK
Filtered MSK
GMSK
(f-f
o
) / Rb 0 1 2 3
-80
-60
-40
-20
0
POWER SPECTRAL
DENSITY
dB
Spectral
characteristics
of GMSK
ECE Department Florida Institute of Technology
Page 43
Tail Traffic/Signaling Flag Training Sequence Flag Traffic/Signaling Tail
3 57 1 26 1 57 3
Sequence used for equalizer training
Equalization
Necessary due to the multipath
propagation
Needs to have :
o Fast convergence
o Low complexity
Two modes of operation
1. Training
2. Equalization
GSM equalizer capable of equalizing for
two equal multi paths separated by 16
microseconds
Introduces overhead of about 18%
RF
Processing
Adaptive
Equalizer
Equalization
Algorithm
Extraction of
Synchronization
Bits
Unequalized
Data
Equalized
Data
ECE Department Florida Institute of Technology
Page 44
GSM Network Features
Mobile Assist Handoff (MAHO)
Discontinuous Transmission (DTX)
Dynamic Power Control (DPC)
Frequency Hopping (FH)
Intercell Handoff
ECE Department Florida Institute of Technology
Page 45
Mobile Assisted Handoff (MAHO)
GSM Implements
MAHO
In the process of
evaluating handoff
candidates, GSM
systems evaluate
measurements
performed by both the
MS and BTS
There are three types
of measurements:
1. Signal Strength
Measurements
2. Signal Quality
Measurements
3. Timing Advance
Measurements
Measurement
type
Link Cell DTX Measurement
Source
RSL Downlink Serving Cell Full Set Mobile
RSL Downlink Serving Cell Subset Mobile
RSL Downlink Neighbors N/A Mobile
Quality Downlink Serving Cell Full Set Mobile
Quality Downlink Serving Cell Subset Mobile
RSL Uplink Serving Cell Full Set BTS
RSL Uplink Serving Cell Subset BTS
RSL Uplink Neighbors Full Set BTS
RSL Uplink Neighbors Subset BTS
Quality Uplink Serving Cell Full Set BTS
Quality Uplink Serving Cell Subset BTS
Timing Advance Uplink Serving Cell N/A BTS
Measurement
type
Link Cell DTX Measurement
Source
RSL Downlink Serving Cell Full Set Mobile
RSL Downlink Serving Cell Subset Mobile
RSL Downlink Neighbors N/A Mobile
Quality Downlink Serving Cell Full Set Mobile
Quality Downlink Serving Cell Subset Mobile
RSL Uplink Serving Cell Full Set BTS
RSL Uplink Serving Cell Subset BTS
RSL Uplink Neighbors Full Set BTS
RSL Uplink Neighbors Subset BTS
Quality Uplink Serving Cell Full Set BTS
Quality Uplink Serving Cell Subset BTS
Timing Advance Uplink Serving Cell N/A BTS
Measurement
type
Measurement
type
Link Link Cell Cell DTX DTX Measurement
Source
Measurement
Source
RSL RSL Downlink Downlink Serving Cell Serving Cell Full Set Full Set Mobile Mobile
RSL RSL Downlink Downlink Serving Cell Serving Cell Subset Subset Mobile Mobile
RSL RSL Downlink Downlink Neighbors Neighbors N/A N/A Mobile Mobile
Quality Quality Downlink Downlink Serving Cell Serving Cell Full Set Full Set Mobile Mobile
Quality Quality Downlink Downlink Serving Cell Serving Cell Subset Subset Mobile Mobile
RSL RSL Uplink Uplink Serving Cell Serving Cell Full Set Full Set BTS BTS
RSL RSL Uplink Uplink Serving Cell Serving Cell Subset Subset BTS BTS
RSL RSL Uplink Uplink Neighbors Neighbors Full Set Full Set BTS BTS
RSL RSL Uplink Uplink Neighbors Neighbors Subset Subset BTS BTS
Quality Quality Uplink Uplink Serving Cell Serving Cell Full Set Full Set BTS BTS
Quality Quality Uplink Uplink Serving Cell Serving Cell Subset Subset BTS BTS
Timing Advance Timing Advance Uplink Uplink Serving Cell Serving Cell N/A N/A BTS BTS
ECE Department Florida Institute of Technology
Page 46
MAHO - Signal Strength Measurements
Performed on uplink and downlink
Reported as a quantized value RXLEV:
RXLEV = RSL[dBm] + 110
Minimum RXLEV:
-110, MAX RXLEV = -47
On the downlink, measurement
performed for both serving cell and up
to 32 neighbors
Up to 6 strongest neighbors are
reported back to BTS through SACHH
Example measurement report
ECE Department Florida Institute of Technology
Page 47
MAHO - Signal Strength Measurements
Measurements of the
neighbors are performed on
the BCCH channels not
affected by the DTX
Measurements on the serving
channel affected by the
DTX.
Perform over a subset of
SACCH that guarantees
transmission even in the case
of active DTX
Before processing, the RXLEV
measurements are filtered to
prevent unnecessary handoffs
-100
-90
-80
-70
-60
-50
-40
0 500 1000 1500 2000
Measurement
R
X
L
E
V
(
d
B
m
)
510
520
530
540
550
560
570
580
B
C
C
H
A
R
F
C
N
RX LEV (dBm) BCCH
Example RSL measurement
ECE Department Florida Institute of Technology
Page 48
MAHO Signal Quality Measurements
Performed on uplink and downlink
Only on the serving channel
Reported as a quantized value RXQUAL
For a good quality call RXQUAL < 3
Measurements are averaged before the
handoff processing
If DTX is active, the measurements are
performed over the subset of SACCH that
guarantees transmission
RXQUAL BER
0 Less than 0.1
1 0.26 to 0.30
2 0.51 to 0.64
3 1.0 to 1.3
4 1.9 to 2.7
5 3.8 to 5.4
6 7.6 to 11.0
7 Above 15
RXQUAL BER
0 Less than 0.1
1 0.26 to 0.30
2 0.51 to 0.64
3 1.0 to 1.3
4 1.9 to 2.7
5 3.8 to 5.4
6 7.6 to 11.0
7 Above 15
RXQUAL RXQUAL BER BER
00 Less than 0.1 Less than 0.1
11 0.26 to 0.30 0.26 to 0.30
22 0.51 to 0.64 0.51 to 0.64
33 1.0 to 1.3 1.0 to 1.3
44 1.9 to 2.7 1.9 to 2.7
55 3.8 to 5.4 3.8 to 5.4
66 7.6 to 11.0 7.6 to 11.0
77 Above 15 Above 15
RXQUAL mapping table
RXQUAL
measurements
Measurement report
ECE Department Florida Institute of Technology
Page 49
Performed on uplink (BTS)
Only on the serving channel
Used by the BTS to estimate
distance to the MS
Expressed in number of bits
of TX advancement
Can be between 0 and 63
TA
MAHO Time Alignment Measurement
ECE Department Florida Institute of Technology
Page 50
Discontinuous Transmission (DTX)
Typical voice activity is around 60%
DTX discontinues transmission during
silent periods
Benefits of DTX
o Uplink:
System interference reduction
Lower battery consumption
o Downlink
System interference reduction
Reduction of the intermodulation
products
Lower power consumptions
Downsides of DTX usage:
o MAHO measurements are less accurate
o Voice quality is degraded due to
slowness of VAD
Mobile station Environment Typical
voice
activity
Handset Quiet location 55%
Handset Moderate office noise
with voice interference
60%
Handset Strong voice
interference (ex. airport,
railway station)
65-70%
Hands free /
handset
Variable vehicle noise 60%
ECE Department Florida Institute of Technology
Page 51
Dynamic Power Control (DPC)
There are three reasons for DPC:
1. Reduction of battery consumption
2. Elimination of near-far problem
3. Reduction of system interference
Power Class GSM (900MHz)
[W]
PCS - 1900 / GSM 1800
[W]
1 20
(1)
1
2 8 0.24
3 5 Not Defined
4 2 Not Defined
5 0.8 Not Defined
Power Class GSM (900MHz)
[W]
PCS - 1900 / GSM 1800
[W]
1 20
(1)
1
2 8 0.24
3 5 Not Defined
4 2 Not Defined
5 0.8 Not Defined
Power Class Power Class GSM (900MHz)
[W]
GSM (900MHz)
[W]
PCS - 1900 / GSM 1800
[W]
PCS - 1900 / GSM 1800
[W]
1 1 20
(1)
20
(1)
1 1
2 2 8 8 0.24 0.24
3 3 5 5 Not Defined Not Defined
4 4 2 2 Not Defined Not Defined
5 5 0.8 0.8 Not Defined Not Defined
(1) Not available commercially
ECE Department Florida Institute of Technology
Page 52
Dynamic Power Control (DPC)
DPC for MS
o Depending on its power class, MS can adjust its power between the max and min
value in 2dB steps
o MS can perform 13 adjustments every SACCH period, i.e., 480ms
o Large adjustments > 24 dB will not be completed before the arrival of new
command
o Commonly implemented as BSC feature. Many vendors are moving it at the BTS
level
DPC for BTS
o Vendor specific
o Based on MAHO reports
ECE Department Florida Institute of Technology
Page 53
Hierarchical Cell Structure (HCS)
Incorporates various cell sizes into
layers of RF coverage
Three common layers:
1. Umbrella cells (HL = 0)
2. Macrocells (HL = 1)
3. Microcell (HL = 2)
HCS provides a way to assign
preference levels between the cells
Very effective way for capacity and
interference management
Signal
Strength
Reselection
Points
Select Micro-Cell
SS_SUFF
Macrocel
Preferred
Micro-Cell
Distance
HL = 1
HL = 2
ECE Department Florida Institute of Technology
Page 54
Handling of Fast Moving Mobiles
If the mobile is moving at a high speed, it will
spend a short time in the coverage area of the
microcell
To prevent excessive handoffs, a temporal
GSM introduces temporal penalty prevents
immediate handoff initialization
If the duration of mobile stay within the
coverage area is shorter than the temporal
penalty, it will never initialize handoff
ECE Department Florida Institute of Technology
Page 55
Frequency Hopping (FH)
FH - multiple carriers used over the course of radio transmission
o There are two kinds of FH:
1. Slow Hopping change of carrier frequency happens at the rate
slower than the symbol rate
2. Fast Hoping carrier frequency changes faster than the symbol
rate
o GSM implements slow FH Scheme
o Carrier frequency is changed once per time slot
o There are two reasons for frequency hopping
1. Frequency Diversity
2. Interference avoidance
ECE Department Florida Institute of Technology
Page 56
Frequency Diversity of FH
Mobile environment is
characterized with small
scale fading
The depth of signal fade
is a function frequency
If two signals are
sufficiently separated in
frequency domain they
fade independently
Frequency diversity gain
diminishes for fast
moving mobiles
ECE Department Florida Institute of Technology
Page 57
Interference Avoidance of FH
FH averages interference
Allows for tighter reuse of frequencies
Increases the capacity of the system
User 1
User 2
User 3
User 4
User 5
f
1
f
4
f
1
f
1
f
1
f
1
f
1
f
2
f
2
f
2
f
2
f
3
f
4
f
1
f
1
f
2
f
3
f
3
f
4
f
1
f
4
f
3
f
1
f
3
f
4
4T T 2T 3T 5T
4T T 2T 3T 5T
4T T 2T 3T 5T
4T T 2T 3T 5T
4T T 2T 3T 5T
ECE Department Florida Institute of Technology
Page 58
Baseband FH in GSM
Each radio operates on a
fixed frequency
The bursts are routed to
individual radios in
accordance to their hopping
sequence
Advantages of baseband hopping
No need to real time retune simpler
radios
More efficient combiners
Disadvantage of baseband hopping
Number of hopping frequencies limited
by the number of radios
TX/RX
TX/RX
TX/RX
Carri er
Freuqnacy
f
1
Combiner
Carri er
Freuqnacy
f
2
Carri er
Freuqnacy
f
n
1
2
n
Bus for Routing
and Switchning
ECE Department Florida Institute of Technology
Page 59
Synthesized FH in GSM
Each radio is hopping in an
independent way
Radios retune real time
Advantages of synthesized hopping:
Set of the hopping frequencies can be assigned in
an arbitrary way
Disadvantage of synthesized hopping:
Need for expensive and lossy combiners
TX/RX
TX/RX
TX/RX
Carri er
Freuqnacy
f
0
,f
1
,...,f
m
Broadband
Combiner
1
2
n
Carri er
Freuqnacy
f
0
,f
1
,...,f
m
Carri er
Freuqnacy
f
0
,f
1
,...,f
m
ECE Department Florida Institute of Technology
Page 60
FH Algorithms
Random Hopping
Implemented in a pseudo random way
Uses one of 63 available pseudorandom sequences
The actual frequency is obtained as a modulo operation with
number of available frequencies in allocation list (FH group)
, , , , , , ,
3 2 1 4 3 2 1
f f f f f f f
, , , , , , ,
3 2 3 4 4 2 1
f f f f f f f
Cyclic Hopping
o Frequencies are used in the consecutive order
o If the radio is performing cyclic FH the order of frequencies in the
sequence goes from the lowest ARFCN to the highest ARFCN
ECE Department Florida Institute of Technology
Page 61
Intracell Handoff
High Interference
Measurement indicates:
o Poor RXQUAL
o Good RXLEV
There is high probability that the call will improve with the handoff to
different carrier within the same cell
To avoid unnecessary handoffs, system introduces maximum number
of intercell handoffs
ECE Department Florida Institute of Technology
Page 62
GSM RF Planning / Design
Link Budget and Nominal Cell Radius Calculation
Receiver Sensitivity
Required C/I ratio
Mobile Transmit Power
Examples of Link Budget
Calculation of a Nominal Cell Radius
Frequency Planning and Reuse Strategies
Frequency Planning Using Regular Schemes
Automatic Frequency Planning
Capacity of GSM Networks
ECE Department Florida Institute of Technology
Page 63
Migration:
1. High speed circuits
switched data
(HSCSD)
2. Packet switched data
(GPRS,EDGE)
3. Integrated packet
services possibly
under different access
scheme (UMTS)
GSM Migration Towards 3G
GSM 2+
9.6 Kb/sec
HSCSD
64 Kb/sec
GPRS
114 Kb/sec
EDGE
384 Kb/sec
UMTS
2Mb/sec
1999
1Q
2000
2Q
2000
3Q
2001
4Q
2002
Timeline
Data Rate
HSCSD - High Speed Circuit Switched Data
GPRS - General Packet Radio System
EDGE - Enhanced Data GSM Environment
UMTS - Universal Mobile Telephone Service
ECE Department Florida Institute of Technology
Page 64
GSM 2+ Data Services
GSMs traffic channel can support the data transfer of a bit rate up to
9.6Kb/sec
o This data rate can be used for:
Short messages
Fax services
E-mail, etc.
o Circuit switched data services
o Not suitable for Internet
Too slow
Too costly (user would pay for the circuit even if there is no
traffic exchanged
ECE Department Florida Institute of Technology
Page 65
High Speed Circuit Switched Data (HSCSD)
HSCSD is using existing GSM organization to provide data
services of a somewhat higher data rates
It can combine several existing traffic channels into a single
connection, i.e., it allows for mobiles multislot operation
HSCSD can be implemented through software upgrades on
existing networks and no hardware upgrades are needed
Seems to be less accepted by the service providers
ECE Department Florida Institute of Technology
Page 66
GPRS is another new transmission capability for GSM that will be especially
developed to accommodate for high-bandwidth data traffic
GPRS will handle rates from 14.4Kbps using just one TDMA slot, and up to
115Kbps and higher using all eight time slots
It introduces packet switching - can accommodate the data traffic
characteristics
General Packed Radio Data (GPRS)
ECE Department Florida Institute of Technology
Page 67
GPRS Network architecture
New type of
node:
GPRS Service
Node (GSN)
BSC
BSC
MSC
VLR
HLR
AUC
EIR
BTS
BTS
BTS
BTS
BTS - Base Station
BSC - Base Station Contoller
MSC - Mobile Switching Center
VLR - Visitor Location Register
HLR - Home Location Register
AUC - Authentif ication Center
EIR - Equipment Identity Register
Um
Interface
A-Bis
Interface
A
Interface
D
C
PSTN
B
B,C,D,E,F - MAP
Interfaces
SGSN
GGSN
SGSN - Service GPRS Support Node
GGSN - Gateway GPRS Support Node
Gn
Interface
Gr
Outside
Packet
Network
ECE Department Florida Institute of Technology
Page 68
GPRS Call routing
SGSN
GGSN
GGSN
SGSN
BTS
BTS
GPRS - PDN
GPRS - PDN
Routing is performed parallel to the GSM network
ECE Department Florida Institute of Technology
Page 69
Packet switched
Upgrades the modulation scheme
o From GMSK to 8-PSK
o Maximum speed ~59 Kb/sec per time slot, ~473.6 Kb/sec for all 8 time slots
o Variable data rate depending on the channel conditions
Defines several different classes of service and mobile terminals
Enhanced Data GSM Environment (EDGE)
EDGE enabled data mobile
ECE Department Florida Institute of Technology
Page 70
Practically achievable data rates
Theoretical rates are constrained by mobile
power and processing capabilities
Most mobiles support less than the maximum
allowed by standard
Practically achievable data rates
ECE Department Florida Institute of Technology
Page 71
UMTS 3G cellular service
Provides data rates up to 2Mb/sec
Possibly standardized as W-CDMA
Universal Mobile Telephone Service (UMTS)
Outline of UMTS
(WCDMA) network