1-‹nº›

The evolution of CDMA can be described by the various protocol revisions

IS-95, Protocol revision 1
Published in July of 1993
Compatibility standard for wideband spread spectrum cellular mobile telecommunications
described the generation of CDMA channels, power control, call processing, handoffs, and
registration techniques

IS-95A, Protocol revision 2

Published in May of 1995, update to messaging and parameters

IS-95A + TSB 74, Protocol revision 3

Published in December of 1995, incorporated PCS CDMA 1900MHz
Improved signaling, added new rate set (Rate Set 2)

TIA/EIA-95-B, Protocol revisions 4 & 5

Published in October of 1998
Included all of protocol revision 3 but removed analog details
Added new capabilities (medium data rate, improvements to handoffs, etc.)

1-‹nº›
IS-2000, Protocol revisions 6 and 7
Provides support for voice and concurrent packet data service with rates as high as 144 Kbps
Standard supports supplemental channel data rates as high as 307.2 Kbps for 1x, and as high
as 1.03 Mbps for 3x

IS-2000, Protocol revisions 9 and 10

Data and Voice evolution for CDMA2000. It supports both data and voice on the same
channel and data speeds of up to 3.1 Mbps can be achieved. The system is defined under
IS-2000. The first offering appears in Release C of the standard where the high data rate is
only available in the downlink, the uplink allows for data at the same rate as 1X. Release D
allows for the high speed data in both directions.

IS-856 (TIA-856)
The initial design of EV-DO was developed to meet requirements for a greater-than-2-Mbit/s
down link for stationary communications, as opposed to mobile communication such as a
moving cellular phone. Originally, 1xEV-DO stood for "1x Evolution-Data Only", referring to
its being a direct evolution of the 1xRTT air interface standard, with its channels carrying
only data traffic.

1-‹nº›
FDMA
FDMA assigns users to specific frequency bands. The user has the sole use of the frequency
band for the entire call duration within the cell. Analog systems use FDMA. In an analog
system, cell sites operating on the same frequency must be separated by a large enough
distance to minimize mutual interference

TDMA
TDMA is an assigned frequency band shared among a few users. However, each user is
allowed to transmit in predetermined time slots. Hence, channelization of users in the
same band in achieved through separation in time. TDMA is a digital technology. As in an
FDMA system, TDMA cell sites with the same frequency must be separated by a large
enough distance to minimize the impact of mutual interference. GSM is a version of TDMA
technology

CDMA
CDMA is a method in which users occupy the same time and frequency allocations, but are
channelized by uniquely assigned codes. The signals are separated at the receiver by using
a correlator that accepts only signal energy from the desired channel. Undesired channels
only contribute to noise. Unlike other technologies, CDMA cell sites share the same
frequency

1-‹nº›
CDMA Concept
The concept of CDMA technology is analogous to communicating while at a party

CDMA “Cocktail Party”

At the party, all people are talking in the same room simultaneously. Imagine that every
conversation in the room is being carried out in a different language that you do not
understand. They would all sound like noise from your perspective. If you knew the
language (analogous to a CDMA code) you could filter out the other conversations and
focus only on the conversation of interest to you

A CDMA system must filter the other conversations (or interference) in the same manner

Even with the knowledge of the appropriate language, the conversation of interest may not
be completely audible. The listener can signal the speaker to increase his volume and can
signal others to decrease their volume. This is analogous to the power control process
used by CDMA systems

1-‹nº›
Digital Functions (X-OR)

X-OR gates are used in cdmaOne coding processes, specifically to spread signals with
orthogonal codes and to generate Psuedorandom Noise (PN) codes

X-OR Gate

The figure above depicts a two-input X-OR gate and its corresponding truth table. A
and B denote the inputs to the gate, while Y represents its output. The X-OR gate
outputs a logic “1” when the two inputs are at opposite levels. The output of the
X-OR gate is zero if the two inputs are at the same level. This is similar to
multiplying two analog voltage levels when the voltage levels are either +1 volt or
-1 volt
A x B = Y
+1 +1 +1
+1 -1 -1
Assume: Logic 0 = +1 volt -1 +1 -1
Logic 1 = -1 volt -1 -1 +1

1-‹nº›
Orthogonal Sequences

Orthogonal sequences are crucial to the CDMA encoding process. Orthogonal sequences are
used on the forward link to uniquely distinguish one subscriber’s traffic from a different
subscriber’s traffic

Orthogonal signals or sequences have zero cross-correlation. Zero cross-correlation is

obtained if the product of two signals, summed over a period of time, is zero. For the special
case of binary sequences, the values 0 and 1 may be viewed as having opposite polarities.
Thus when the product (X-ORing in this case) of two binary sequences results in an equal
number of 1s and 0s, the cross-correlation is zero

Properties of orthogonal sequences

• zero cross-correlation
• equal number of 1s and 0s (except the all 0s sequence)
• if X-ORed with another orthogonal sequence, the result is an orthogonal sequence

1-‹nº›
Correlation (cont.)

The example above shows two different signals. However, the two signals are uncorrelated
and therefore knowing one of them does not provide any information of the other

1-‹nº›
Correlation

Correlation is a measure of similarity between any two arbitrary signals. It is computed by

• multiplying the two signals
• then summing (integration) the result of a defined time window
• lastly, dividing the result by the total number of clock cycles

If the result is “1” the two signals have 100% correlation and are said to be identical. An
example of this is shown in the figure above

1-‹nº›
Generating Orthogonal Sequences

Orthogonal sequences are easily generated by starting with a seed of 0, repeating the 0
horizontally and vertically, and then inverting to a 1 diagonally. This process may continue
with the newly generated block until the desired codes with the proper length are generated

Sequences generated in this manner are referred to as “Walsh Codes”

In the above illustrated example, we start with a seed of 0. Using the process described
above, we will generate four 4-bit Walsh Codes:
0000
0101
0011
0110

1-‹nº›
Walsh Codes

The orthogonal sequences currently used in terrestrial cdmaOne systems are Walsh Codes of
length 64. These Walsh Codes are used in the forward link to separate users’ traffic from one
another. In any given sector, each forward channel is assigned a distinct 64-bit Walsh Code

1-‹nº›
Walsh Code Spreading

Walsh Code spreading is done on the forward link to separate users’ traffic from one another
Data is spread in the RBS by X-ORing the data symbol-by-symbol with a particular Walsh Code.
When the subscriber device receives the spread signal it again X-ORs the signal with the very
same Walsh Code, thus retrieving the original data. This is known as despreading. The
process can be expressed logically as follows

Assumptions

A 0=A
Logical equation for spreading and despreading data
A A=0

Where A is the original data, and B is the Walsh Code used for spreading in the RBS and
(Adespreading
B) B the
in = Asubscriber
(B device
B) = A 0=A

1-‹nº›
Spreading with an Orthogonal Function
Spreading the user data (symbols from the FEC) is done by X-ORing each symbol with an entire
walsh code

Orthogonal Spreading

Each symbol is X-ORed with all the chips in the orthogonal sequence (Walsh Code) assigned
to the user. The result is then processed and transmitted over the physical channel after
being combined with the other forward link signals

In the above example the user data is spread with a 4-bit Walsh Code. The product of the
user symbols and the 4-bit Walsh Code is a sequence of digits that must be transmitted at
four-times the rate of the original user data sequence

Note that this is a simplification used to exemplify the process; 64-bit Walsh Codes are used
for spreading in cdmaOne

1-‹nº›
Recovery of Spread Symbols using the Correct Walsh Code
In order to recover the original user data (symbols from the FEC), the receiver must X-OR the
received signal with the same Walsh Code that was used for spreading

Signal Recovery
The receiver despreads the chips by using the same Walsh Code used at the transmitter. If
there is no noise, the symbols (or digits) are completely recovered without any error. In
reality, the channel is not noise free, but cdmaOne systems employ FEC techniques to
combat the effects of noise and to enhance the performance of the system

The above example shows the the received data being X-ORed with the same 4-bit Walsh
Code used at the transmitter. The result is the original user data (symbols)

Note that this is a simplification used to exemplify the process; 64-bit Walsh Codes are used
for despreading in cdmaOne

1-‹nº›
Recovery of Spread Symbols Using the Wrong Walsh Code
If the wrong Walsh Code is used to despread the signal received at the transmitter, the result
is a set of digits that provide no information of the original user data

Signal Recovery

When a different Walsh Code is used for despreading, the resulting set of digits yield an
equal number of 1s and 0s, and thus the original digit cannot be determined. This is a
clear demonstration of the advantage of the orthogonal property of Walsh Codes.
Whether a different sequence is used by the target user or other users attempting to
decode the received signal, the result always has an equal number of 1s and 0s

1-‹nº›
Spreading with Multiple Signals
On the forward link, the individual spread signals are combined in the RBS and transmitted as
one signal on the same frequency
Example of combining a three spread signals
The illustration above shows three signals being combined into one signal for transmission
User A
Original data from FEC: 00
Walsh Code used for spreading: 0101
User B
Original data from FEC: 10
Walsh Code used for spreading: 0011
User C
Original data from FEC: 11
Walsh Code used for spreading: 0000

The analog signal shown on the bottom of the slide is the composite signal when the spread
signals of all three users are summed together

1-‹nº›
Despreading a Composite Signal
The composite signal is received by the subscriber device on the forward link. The subscriber
device must despread the composite signal with the correct Walsh Code to retrieve its specific
user data

Example of despreading a the composite signal

The subscriber device receives the composite signal and despreads it with the Walsh Code
assigned to User A to retrieve User A’s data. The composite signal is X-ORed with user A’s
assigned Walsh Code, and the result is then averaged by according to the bit size of the
Walsh Code. In the example above, we are using 4-bit Walsh Codes, though cdmaOne
spreads and despreads using 64-bit Walsh Codes. If the correct Walsh Code is used, the
result of the averaging will determine the original data of User A regardless of the number
of combined signals that were sent from the RBS on the forward link. Again, this is due to
the orthogonal properties of Walsh Codes

The original data of user B and user C can be retrieved in the same manner by despreading
with the appropriate user Walsh Code

1-‹nº›
1-‹nº›
CDMA Modulation Overview
cdmaOne systems convert analog voice signals into digital signals for transmission. There are
several steps in the digital transmission process. Many of these steps are common to digital
wireless schemes. After each step in digital processing, the signal conveys a different meaning
and several terms are used to refer to the signal at different stages in the process. Bits,
symbols, and chips all look the same: a single binary digit. What distinguishes one from
another is their relationship to the stage of processing
Bits
A bit is the fundamental unit of information; a single binary digit. Analog information is
encoded into a sequence of binary digits (A/D conversion). Both user data and error
detection code digits are considered “bits”. The bit rate is the measure of the volume of
information being transmitted
Symbols
A symbol is the output of the coding process known as the Forward Error Correction (FEC).
Each incoming bit produces several code symbols. The symbol rate is a measure of the
redundancy introduced by the FEC scheme
Chips
The output digits of a spreading code generator are commonly termed “chips”. Several chips
are used to spread a single code symbol. The chip rate is a measure of the amount of of
spreading performed

1-‹nº›
Analog to Digital Conversion (A/D conversion)
Elementary analog-to-digital conversion is commonly referred to as “Waveform Coding”. The
simplest method of waveform coding is one that directly samples the speech signal, quantizes
the sample, and then encodes each sample into a binary code. This process produces a Pulse
Codes Modulated (PCM) signal with a data rate of 64 Kbps for voice telephony. This process
occurs at the PSTN on the forward link, and it occurs at the subscriber device on the reverse
link
Low Pass Filtering (LPF)
The analog signal is filtered to allow only frequencies below 4 KHz to pass
Sampling
The samples of the analog signal must be taken at twice the highest possible frequency.
Since the LPF allows a maximum frequency of 4 KHz to pass, sampling is done 8000 times
per second
Quantizing
Quantization is the process of assigning an 8-bit value (for PCM) to each of the samples. This
gives a dynamic range of 256 different values
Encoding
Encoding is a compression process converting the digitized signal from a linear form to a
logarithmic scale. The conversion insures that low-amplitude signals have a minimal loss of
fidelity. The signal is later expanded at the receiver

1-‹nº›
Voice Coder (Vocoder)
Voice coders (Vocoders) compress the high rate digital voice signal from 64 Kbps (PCM) down
to between 4 and 13 Kbps depending on the technique used. Vocoders take advantage of the
pauses in human speech patterns to compress the digital signal. This process occurs in the BSC
on the forward link and occurs in the subscriber device on the reverse link

Linear Digital Conversion

The digital PCM speech signal is transcoded from the compressed 8-bit per sample
logarithmic form into a 13-bit per sample linear form. The conversion to 13-bits per
sample provides the system with better dynamic range. The signal is then sent into the
Code Excited Linear Prediction (CELP) encoder

Code Excited Linear Prediction (CELP)

The CELP encoder outputs a variable rate frame of compressed voice every 20 msec. The
rate of the digitized compressed voice in the frame depends on voice activity and will be
one of four different rates (for the service option selected). The average rate is roughly 4
Kbps or 7 Kbps depending on the service option selected

1-‹nº›
Error Detection Codes
Each variable rate vocoded frame contains a number of additional check bits used to detect
errors when received at the destination. These additional check bits are known as Cyclic
Redundancy Checking (CRC) bits

Cyclic Redundancy Checking (CRC)

CRC provides a high level of error detection. The price paid for this detection capability is
overhead bits added to the information bits. these overhead bits are appended to the
information bits prior to transmission to the FEC. Within each 20 msec variable rate
vocoded frame there may be as many as 12 CRC bits and as few as 0 bits

1-‹nº›
Variable Rate Vocoding
The vocoder compresses the output of the A/D converter to a lower bit rate. This saves
valuable bandwidth on the backhaul between the BSC and the RBS. The variable rate vocoder
takes advantage of low speech activity and transmits a lower rates, thereby reducing the
average transmission rate to roughly 4 or 7 Kbps

Examples of different rates

The example above illustrates the vocoder outputting variable rate frames at
• full rate (192 bits)
• half rate (96 bits)
• quarter rate (48 bits)
• eighth rate (24 bits)

Each of these frames are transmitted every 20 msec, and the vocoder is utilizing Radio
Configuration 1

1-‹nº›
Radio Configurations
CDMA2000 systems can employ different Radio Configurations according to the standard. The
Radio Configurations supported in CDMA2000 initial systems are shown above. The frame
will contain user data or digitized, compressed voice information at one of the data rates
shown above. The data rate depends on the level of voice or data activity. Note that data
bypasses the vocoding (voice compression) process since the information is not voice
information.

Radio Configuration
A “Radio Configuration” (RC) is a set of traffic channel formats. Each RC is comprised of
different bit rates. A RC may carry voice, user data, and/or signaling depending upon the
code channel and service option. The transmit rate of the frame changes based on the
voice activity or based upon the bandwidth required for a given amount of data.

1-‹nº›
Forward Error Correction (FEC)
Voice transmissions require “real time” transmission. Delays are unacceptable to the human
ear. It is impractical to attempt to retransmit a portion of a conversation that arrives at the
receiver in error. For this reason, the signal must be transmitted reliably. One method to
increase reliability is to simply transmit at higher power (not good for CDMA). An alternate
method is to use some form of coding that enables the receiver to correct some errors. FEC
occurs in the RBS on the forward link, and it occurs in the subscriber device on the reverse link
FEC Coding
Coding that enables the receiver to correct errors is referred to as Forward Error Correction
coding. A very unsophisticated form of coding is illustrated above. Redundancy is
introduced simply by repeating the digits. In cdmaOne, we use more sophisticated
techniques introduce redundancy through other means
Interleaving
The purpose of interleaving is to enable the recovery of information when burst errors occur
(i.e., loss of consecutive symbols over the RF channel)
Interleaving jumbles the symbols before transmission over the RF channel. The receiver puts
the data back to the original order (deinterleaving), thereby uniformly distributing the
impact of the burst error across the frame. The receiver can then use a simple majority
logic decoder to determine the original sent digits

1-‹nº›
Turbo Encoder

Data transmissions may require more bandwidth than voice, and the “realtime” consideration
(in some instances) is no longer as important as with voice. As will be shown later, higher
bandwidth is achieved using shorter Walsh Codes. To help compensate for the shorter Walsh
Codes, turbo coding provides more data protection than the coding used with voice
transmissions. However, the complexity of the turbo coding may cause processing delays
during the decoding process. Therefore, turbo coding is best suited for non-realtime data
transmissions but not well suited for realtime data and voice transmissions.

Turbo encoding may be used for

data rates of 19.2 kbps or greater.

The adjacent table shows

the error correction
improvement (in terms of
dB) when compared
to the Convolutional Encoder.

1-‹nº›
Spread Spectrum
A system is defined to be a spread spectrum system if it meets the following:
• The transmitted signal occupies a bandwidth much greater than the minimum bandwidth
needed to send the information
• Spreading is accomplished by means of a “spreading signal”, often called a “code signal”,
which is independent of the data
• At the receiver, despreading (recovering the original data) is accomplished by the
correlation of the received signal with a synchronized replica of the spreading signal

Why spread?

In 1948, Claude Shannon proved that error-free communication is possible on a noisy

channel provided that the data rate is less than the channel capacity. His equation
(“Shannon’s Equation”) showed that channel capacity depends linearly on the bandwidth
and logarithmically on the the signal-to-noise (S/N) ratio

Therefore, larger gains in capacity are realized by increasing the bandwidth than by
increasing the S/N ratio. Furthermore, increasing bandwidth allows a much lower S/N
ratio (or power) for a constant capacity

1-‹nº›
Spread Spectrum Techniques
There are different methods for spreading a signal over a defined bandwidth. cdmaOne uses a
technique known as “Direct Sequenced” spreading. This occurs in the RBS on the forward link,
and it occurs in the subscriber device on the reverse link

Direct Sequenced spreading

The voice information signal is inherently narrowband, on the order of less than 10 KHz. The
energy from this narrowband signal is spread over a much larger bandwidth by multiplying
the information signal by a wideband spreading code. In cdmaOne, the chip rate of the
spreader is 1.2288 Mcps (Mega chips per second)

1-‹nº›
Channel Encoding Process
The processes for encoding a forward link code channel and for encoding a reverse link code
channel is very similar, and can be described by three major processes: Error Correction,
Channel Separation, and Transmitter Separation. The encoding processes used for Channel
Separation and Transmitter Separation will be discussed in this section.

Error Correction
This process allows real-time burst errors incurred during transmission to be removed at the
receiver.

Channel Separation
This process uniquely encodes each code channel sent from a single transmitter. This allows
the receiver to separate the channels as they are received.

Transmitter Separation
This process tags each transmitter with a specific code sequence. This allows the receiver to
distinguish between signals from different transmitters.

1-‹nº›
Pseudorandom Noise (PN) Sequences
PN sequences are used in both the forward and reverse links. Later it will be shown that these
sequences can be manipulated to provide very specific information about the transmitting
source (RBS sector on the forward link, subscriber devices on the reverse link). For example, a
manipulation of a PN sequence on the forward link allows the subscriber device to identify the
specific transmitting RBS sector. A manipulation of a PN sequence by each subscriber device
on the reverse link allows the RBS to distinguish the different subscriber devices from one
another
Properties of PN Sequences
A pseudorandom sequence is a deterministic sequence of 1s and 0s that mimic randomness
properties of a Bernoulli sequence, sometimes referred to as a “coin-flipping” sequence.
Each 1 and 0 corresponds to a “heads” or “tails” outcome in a succession of independent
coin-flip experiments
PN Sequence Generation
The state machines generating PN codes are very simple and consist of shift registers and X-
OR gates. The length of a PN code is determined by counting the number of bits in the
sequence before it begins to repeat itself. In general, if the number of shift registers is
represented by “r”, then the length of the PN code (N) is equal to 2r-1
In the above example, N = 23-1 = 7. Therefore, the PN sequence will repeat every 7 bits

1-‹nº›
Channel Separation
Each sector of an RBS transmits multiple code channels simultaneously on one frequency.
Similarly, a subscriber device may transmit multiple code channels simultaneously on one
frequency. Therefore, CDMA2000 technology must have a method for separating these code
channels from each other at the receiver.
Forward link channels are separated from each other by applying a unique Walsh Code to each
channel. These Walsh Codes are unique within a sector, but these same Walsh Codes are used
in every sector. Sectors are separated using unique offsets of the Short PN sequences (as will
be seen later in this section).

Forward Link Code Channels

The graphic above shows the forward link code channels that will be supported by initial
CDMA2000 systems. These codes channels will be discussed in detail in the next section,
but they are shown here to illustrate Walsh Code assignments in the forward link.

1-‹nº›