THE MUOS-WCDMA AIR INTERFACE

John S. Sadowsky
David K. Lee
General Dynamics C4 Systems
Scottsdale, AZ

ABSTRACT sec. This requires a rather large dedicated physical

control channel (DPCCH) power overhead to support
WCDMA4 is a 3rd generation wireless communication layer 1 signaling and receiver tracking algorithms. The
system developedfor terrestrial cellular systems. The much longer MUOS UHF coherence times makes this
Mobile User Objective System (MUOS) uses WCDMA4 as costly high DPCCH power overhead unnecessary, and,
a basis for its waveform definition, thus exploiting the furthermore, there is a strong desire to minimize control
extensive development of this advanced commercial channel overhead in order to maximize data bearing
communications technology. MUOS is, however, a geo- capacity. Thus, the MUOS waveform was designed to
satellite system which in certain ways is a vastly operate with significantly less DPCCH and CPICH
different communications environment than the power than is typical in terrestrial WCDMA4. In the
terrestrial cellular environment. This paper shall user-to-base (U2B) direction, the DPCCH (which
overview key changes to the WCDMA4 Layer ] to yield included dedicated pilot symbols and physical layer
the MUOS waveform definition. signaling) power is reduced roughly 7 dB. On the B2U
direction, the common pilot channel (CPICH) may be
INTRODUCTION reduced as much as 1U dB. The reduction in DPCCH
power (in both directions) requires a re-design of
The Mobile User Objective System (MUOS) is a geo- physical layer control channel FEC coding.
satellite UHF communications system that utilizes the
WCDMA air interface developed by the 3rd Generation In order to achieve time diversity gains with the very
Partnership Project (3GPP) for terrestrial cellular long coherence times, the MUOS coding system shall
systems. The geo-satellite link, however, poses a employ a Dovetail-Interleaving (DTI) scheme. In
number of different characteristics than terrestrial comparison to the typical 3GPP interleaving span of 40
cellular. These include -600 ms round trip propagation m sec, MUOS DTI achieves spans of 640 m sec. This is
delay, tight multipath delay spreads (due to satellite done with a relatively minor augmentation of 3GPP
elevation), and long coherence time (due to a UHF rather interleaving.
than S-band carrier). The satellite footprint is segmented
into 16 beams. Adjacent satellite beams are somewhat THE WCDMA RADIO ACCESS BEARER
analogous to interfering base stations in the terrestrial
case; however satellite beams fade in unison as opposed The Layer 1 and 2 transmit data flow is illustrated in
to independent fading seen in terrestrial systems. Thus, Figure 1 for a WCDMA radio access bearer (RAB).
MUOS cannot exploit many of the diversity techniques The inter-layer interface consists of logical channels
designed into 3GPP WCDMA. called transport channels (TrCH). Typically there is one
TrCH for higher layer control, and there are one or more
This paper is a survey of the various MUOS changes to TrCHs for user dedicated data. TrCH data is presented
the 3GPP WCDMA Layer 1 definition. An important as transport blocks (or simply "blocks"). A TrCH
topic for the full understanding of the MUOS waveform transport format is the number of blocks to be
is power control. This paper shall discuss how the transmitted. (In 3GPP WCDMA, block size is also a
power control feedback is encoded and communicated variable transport format parameter. Each MUOS TrCH
over physical control channel. We do not, however, has a single fixed block size.)
discuss any details of the power control algorithm -
which is a topic worthy of an entire paper of its own. Layer 2 has two sub-layers: radio link control (RLC)
and media access control (MAC). There is one
3GPP WCDMA was designed to deal with S-band instantiation of RLC per TrCH. It performs packet
Doppler that can result in coherence times less than 1 m segmentation and transport block construction, queuing

1-4244-1513-06/07/$25.00 ©2007 IEEE I of 6

and ARQ. The list of Transport Formats, one from each exactly match the number of TrCH coded bits to the
TrCH, is called a transportformat combination (TFC). allocated number of DPDCH bits.
The main job of the MAC, in a nutshell, is to examine
the blocks ready to transmit in the RLC queues and The lower sub-layer of Layer 1 is labeled the modulation
select the TFC. MAC TFC selection is constrained by sub-layer in Figure 1. This sub-layer performs
TrCH priority and (in the U2B) available Tx power. modulation, spreading and scrambling. WCDMA
Each RAB has a legal list TFCs - not all combinations utilizes Orthogonal Variable Spreading Factor (OVSF)
are allowed. The MAC communicates its selection to codes for spreading data, and a PN code for scrambling.
Layer 1 via the TFC indicator (TFCI). The TFCI is an The chip rate is 3.84 Mcps. After spreading and
important Layer 1 control signal that must be received scrambling the waveform is constructed using a root
and correctly decoded at prior to de-interleaving and raised cosine chip waveform.
decoding of dedicated data.
Layer 1 outputs DPDCH data on a 10 m sec time
interval called aframe. The Layer 1-2 interface operates
once per Transmission Time Interval (TTI), which may
be 1, 2, 4 or 8 frames. Most MUOS RABs have an 80
IT L ms (8 frame) TTI. 3GPP allows different TrCHs to have
yer 2 Transport Blocks
La rufl TrCH ....per TTI different TTIs within the same CCTrCH, but multiple
yer 1 TTIs are not allowed in MUOS. TTI is thus a key
parameter for determining the channel coding time span,
U-
DPDCH data which is important for time diversity.
per frame
Frames are divided into 15 slots. The slot duration is 2/3
I/
A m sec. In 3GPP WCDMA, the slot division provides the
DPCCH Modulated 1500 bps transmit power control feedback rate. Clearly,
DPCH frames due to the long round trip delay, MUOS cannot support a
fast (one bit per slot) feedback rate. MUOS TPC
Figure 1: The Radio Access Bearer. feedback is transmitted via a new control channel called
the physical control channel (PCC). Nonetheless, the
Layer 1 constructs the dedicatedphysical channel slot interval is important for MUOS DPCCH
(PDCH), which has two parts: the dedicated physical construction.
data channel (DPDCH) that consists of coded bits from
the Coded Composite Transport Channel (CCTrCH), Both TFCI and PCC are encoded using the 3GPP (32,10)
and the dedicated physical control channel (DPCCH), Reed-Muller code, in separate code words.
which consists of the TFCI (from the MAC) and the
PCC (power control feedback bits). MUOS does not implement the 3GPP compressed mode
(which is primarily used for inter-frequency and
As shown if Figure 1, Layer 1 is partitioned into two WCDMA -* GSM handover measurements).
sub-layers. The upper sub-layer is the CCTrCH that
performs the following functions separately for each THE USER-TO-BASE WAVEFORM
TrCH: CRC attachment (to each transport block), FEC
(forward error correction) channel coding (K9 Waveform Construction
convolutional or rate 1/3 PCCC turbo), rate matching, Ist
interleaving and frame segmentation. The coded and The MUOS User-to-Base (U2B) waveform is very
interleaved data from the different TrCHs are then similar to 3GPP. DPDCH and the DPCCH are BPSK
multiplexed into coded composite DPDCH data frames, modulated, spread via orthogonal OVSF codes and I-Q
and a 2nd interleaver mixes data across TrCHs on a per multiplexed. The DPDCH is on the I-channel and
frame basis. DPCCH is on the Q-channel. After I-Q multiplexing,
the composite symbols are scrambled using a QPSK
Rate matching is an important algorithm. It operates on scrambling codes as described in [3]. Each UE is
a per TrCH basis, but is coordinated across TrCHs. Rate assigned a distinct scrambling code, which differentiates
matching allocates DPDCH bits between the TrCHs, and UE signals at the base station receiver.
then applies puncturing or repetition (after channel
coding), and DTX bit insertion in the B2U, in order to

2 of 6
The DPCCH Spreading Factor (SF) is 256, resulting in terrestrial multipath than that occur from ground
ten DPCCH symbols per slot. Of these ten, six are propagation in a terrestrial cellular system. While
pilots, two are coded TFCI and two are coded PCC bits, MUOS does support some high velocity vehicles, such
as discussed below. as supersonic aircraft, these are generally not embedded
in a rich multipath environment. Aircraft velocity
The DPDCH SF varies from 256 to 8, in powers of 2, results in a large Doppler shift, not a Doppler spread that
depending on data rate. SF is selected by the CCTrCH would result in a reduced coherence time. MUOS
for each TFC, and the resulting number of DPDCH bits vehicles that are embedded in a terrestrial multipath
is 10x256/SF per slot, or 150x256/SF per frame. For environment generally do not have the high velocities
each TFC, the SF is selected so that rate matching is expected in terrestrial cellular systems. The end result is
achieved with no more than 100% coded bit repetition. that MUOS coherence times are typically on the order of
(The U2B rate matching algorithm strongly discourages 100 m sec or longer.
puncturing in the U2B direction.) The MUOS U2B rate
matching algorithm is the same as that used in 3GPP; see MUOS channels do experience ionospheric scintillation.
[2]. About 85% of the users are expected to experience
"optimal" ionospheric conditions, about 1% are
The DPDCH/ DPCCH power ratio is also TFC "severely stressed," and the rest are "weakly stressed."
dependent (in proportion to data rate) as in 3GPP. As in Optimal and weakly stressed channels have a strong
3GPP, power control controls DPCCH power, while Ricean line-of-sight component; strongly stressed
DPDCH power is determined by the TFC dependent channels are purely Rayleigh. Stressed ionospheric
power ratio, which can change on TTI boundaries. scintillation also has a coherence time, but again, it is
typically on the order of 200 m sec or longer.
The MUOS U2B DPCCH has 6 pilot symbols, 2 coded
TFCI and 2 coded PCC bits per slot. The 32 bit 3GPP While most MUOS channels benefit from a relatively
Reed-Muller code is punctured to 30 bits, 2 of which are strong line-of-sight Ricean component, unfortunately
transmitted per slot (as in 3GPP [2]). Recall that in MUOS does not benefit from the various diversities
3GPP different TrCHs can have different TTI, which present in terrestrial cellular. The MUOS satellite
means that the TFCI can change on frame boundaries. footprint is segmented into 16 beams, and the base
This is why the TFCI is encoded and decoded on each station receiver can perform beam combining - which is
frame in 3GPP. In MUOS, however, TFCI is constant similar in implementation to terrestrial cellular antenna
across the (typically 8 frame) TTI. This allows the diversity (typically referred to as "softer handover")
receiver to accumulate TFCI symbols across the TTI between sectors. However, all beams see the same
prior to TFCI decoding. Repetition across frames terrestrial/inonspheric fading channel. Thus, beam
significantly improves TFCI reliability. For an 8 frame combining is really a beam forming technique that
TTI the repetition gain is about 9 dB, which overcomes focuses a beam on the UE to reduce multiple access
the 7 dB reduction (as mentioned in the introduction) of interference (MAI). It is not a true diversity technique.
DPCCH power level relative to a terrestrial system.
The only true diversity that is available to the MUOS
Likewise, the U2B PCC (power control feedback) is U2B link is time diversity. However, as explained
transmitted once every 16 frames. The 30 bit Reed- above, MUOS coherence times can be on the order of
Muller coded PCC is repeated across these frames and 100 m sec or longer, while the maximum TTI is only 80
decoded once after accumulating up to 16 frames of m sec. Thus, it is desirable to significantly increase the
repeated PCC. interleaving time span without breaking the 3GPP
CCTrCH flow.
Time Diversity Enhancement via Dovetail Interleaving
The solution is dovetail interleaving (DTI). Recall in
As noted in the introduction, the MUOS UHF channel Figure 1 that the CCTrCH encodes transport blocks once
fading characteristics are significantly slower than the per TTI, while the modulation sub-layer of Layer 1
terrestrial cellular S-band case. All else equal, operates frame-by-frame. DTI is a frame interleaver that
coherence times are roughly 6 are longer at UHF is inserted between the two Layer 1 sub-layers, as shown
compared to S-band. However, all is not equal. in Figure 2.
Coherence time, which is inversely related to Doppler
spread, results from the UE's velocity relative to a dense
multipath environment. Satellite elevation results in less

3 of 6
 TFCI BLER for 9.6 kbps RAB
1 E+00

DPDCH frames
1E-01
Interleaved
DPDCH frames
plus TFCI m

PCC 1 E-02

Figure 2: DTI enhanced Layer 1.

1 E-03
The operation of DTI is shown in Figure 3. Data from 2 3 4
Eb/NO (dB)
5 6 7 8

successive TTIs are encoded by the CCTrCH and

labeled A, B, C, In this example the TTI is 8 frames.
Figure 4: Some DTI performance results.
...

The CCTrCH encoded DPDCH frames are labeled AO,

.., A7 for TTI A, BO, .., B7 for TTI B, and so on. The
dovetail interleaver receives all 8 encoded frames of a
Figure 4 presents some sample DTI performance results
for the 9.6 kbps MUOS RAB on three different MUOS
single TTI at once, and stores most in memory. It then channel models. This RAB may be operated with DTM
plays them out over time, interleaved with DPDCH = 8 or with no DTI (DTM = 1). The results show 2-2.5
frames from other TTIs in a "dovetail fashion." The dB improvement on stressed channels, and almost 1.5
total transmission time for a given TTI's data is DTM x
dB improvement on one optimal channel.
TTI, where DTM is the dovetail multiplier. DTM = 1
indicates no DTI. In Figure 3, DTM = 4.
mmmmmr-- r
TTIB TTIC mD
THE BASE-TO-USER WAVEFORM

The B2UDPCCH

The Base-to-User (B2U) DPCCH consists of TFCI and

PCC - there are no dedicated pilot symbols (unlike the
Order of Tx 5 i
3GPP downlink). As in 3GPP, the DPCCH and DPDCH
Tx Time = DTM x TTI are both QPSK modulated and time multiplexed within
each slot.
Figure 3: DTI operation on frames.
B2U MUOS beams are like sectors in a terrestrial
Most MUOS U2B RABs have TTI = 80 m sec and DTM cellular system. All links transmitted on a common
= 8, resulting in an interleaving time span of 640 m sec. beam are scrambled with the same scrambling code, and
Recall that the geo-satellite round trip delay is about 600 various links within a beam are distinguished via
m sec. Thus, DTI complement power control. Fading different orthogonal OVSF codes. Thus, an important
that occurs on a faster than 600 m sec time scale can be issue in the B2U (as in cellular CDMA and WCDMA) is
mitigated using the time diversity that DTI provides. conservation of the OVSF code space.
Fading that occurs slower than 600 m sec time scale can
be mitigated by power control. Unlike the U2B, B2U SF is constant (for the RAB)
across all TFCs. SF is set to support the most heavily
loaded TFC. Rate matching may employ some
puncturing in order to avoid SF reduction, and hence,
conserve OVSF code space. Since the SF is fixed, the
total number of DPCH bits (which includes both
DPDCH and DPCCH bits) per frame is 300x256/SF.
Rate matching is discussed in more detail below.

4 of 6
DPCCH and DPDCH have the same SF. In 3GPP, each balancing block error rate (BLER) across TrCHs on the
slot has the same number of DPCCH bits, and moreover, most heavily loaded TFC. For under-loaded TFCs
the TFCI can be decoded on a frame-by-frame basis. (having fewer transport blocks), the rate matching
(Recall that different TrCHs can have different TTIs in algorithm inserts DTX bits in the DPDCH in order to
3GPP, but not in MIJOS.) For MUOS, this is an maintain the same TrCH puncturing or repetition rates
excessive DPCCH overhead, especially for low data rate across all TFCs. Unlike the U2B where
(hence large SF) RABs. MUOS supports a 2.4 kbps DPDCH/DPCCH power ration varies with TFC, the
RAB at SF = 512. Using the 3GPP design would result B2U symbol amplitude is constant across TFCs. Thus,
in as much as 400 o DPCCH overhead for that RAB. in the B2U power is throttled up or down with data rate
via DTX bit insertion.
Thus, to reduce DPCCH overhead, TFCI and PCC bits
are distributed over the TTI (for TFCI) or PCC cycle, This works fine for a terrestrial system, but not for a
and decoded once. Unlike 3GPP, a frame is constructed military system with strict user data rate requirements.
using three slot formats: T, P and D. A T-slot contains In a terrestrial system, the "advertised" data rate is just
TFCI bits and DPDCH bits, a P-slot contains PCC bits the maximum rate achieved by the most heavily loaded
and DPDCH bits, and a D-slot contains only DPDCH TFC and does not account for block errors or ARQ
bits. Figure 5 shows how TFCI and PCC bits are retransmissions. For example, consider a RAB with one
distributed across frames for a hypothetical 4 frame TTI. dedicated data TrCH and one control TrCH. The TFC is
The first 2 slots of each frame are T-slots. Since there indicated as { ND , NC } where ND is the number of
are 32 Reed-Muller coded TFCI bits, each T-slot must
bear at least 4 TFCI bits. There may be integral order data blocks and NC is the number of control blocks. An
repetition, so the T-slots may have of 4 TFCI bits. "advertised" data rate of 32 kbps might achieved by
Including repetitions, the entire TFCI is distributed transmitting ND= 4 blocks of 320 bits in a 40 m sec
across the frames of the TTI, and at the receiver decoded TTI. The actual data rate that the user sees, however, is
once per TTI. less due to block errors. Nonetheless, the rate matching
rates (puncturing or repetition) will be set by the most
Frame
Number
Slot Number heavily loaded TFC, which might be the {4,1 } TFC.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Alternatively, the {4,1 } combination may be illegal (per
AL0 the RAB defined legal TFC list), so the most heavily
- loaded may be either {4,0} or {3,1 }. It is usually the
2 case that the control TrCH has a higher MAC priority
3 than does data. So control blocks transmission preempt
4 data, and preemption for control blocks is not accounted
__5 for in terrestrial "advertised" data rate.
6
Military systems, such as MUOS, are required to deliver
a guaranteed user data rate. Thus, a MUOS RAB must
include TFCs that transmit data at a higher rate than the
Figure 5: DPCCH insertion in the B2U. specified rate in order to allow for block errors,
retransmissions and periodic control preemption. In the
The B2U PCC is encoded using a (16,5) bi-orthogonal above example, we will need to add {5,0} TFC extended
code, which is a sub-code of the (32,10) Reed-Muller data rate TFC, and probably also the {4,1 } TFC to
code that can be encoded and decoded using the same prevent control preemption at the specified 32 kbps rate.
algorithms. In the example of Figure 5, the PCC These are called extended rate TFCs. (Like the rock
codeword is transmitted over 8 P-slots, with each P-slot band Spinal Tap, our guitar amps with volume knobs
bearing 2 PCC bits. The B2U PCC is transmitted every that go up to 11 instead of just 10.) This presents a
80 m sec, so the four PCC code words shown in Figure 2 problem. The most heavily loaded TFC will be one of
are repetitions. these extended rate TFCs, which sets the puncturing or
repetition rate for all TFCs. Since it is highly desired to
Base-to-User Rate Matching constrain SF in order to conserve OVSF codes, the
A second significant deviation from the 3GPP is in the
extended rate TFCs are very likely force excessive
B2U rate matching algorithm. In the 3GPP algorithm, a puncturing, and then. Excessive puncturing has a
detrimental impact on performance, especially beyond
puncturing or repetition rate is set for each TrCH by

5 of 6
about 20% puncturing for the 3GPP rate 1/3 turbo code.

T
DPCCH amplitude ratio Nominal Set
Amplitude
Once 3GPP sets the puncturing rate for the most heavily
loaded TFC, it maintains this rate for all TFCs. That's a) Extended Rate
TFC Amplitude
OK for terrestrial systems because the most heavily
F-E
loaded TFC is the 'work horse' that delivers the x
advertised data rate. For MUOS, however, we would
have the situation that an infrequently used extended rate
TFC constrains the performance across the entire RAB, T-Slot >< P-Slot P-Slot D-Slot

which is not acceptable.

Figure 3: B2U amplitude profile across one frame.
The solution is a MIJOS modification to the 3GPP rate
matching algorithm. The TFC set is partitioned into two SUMMARY AND CONCLUSION
subsets: the nominal set and the exception set. The
exception set contains the extended rate TFCs. This paper has reviewed the MUOS Layer 1 air interface
as a modification of the 3GPP WCDMA definition. The
The nominal set contains all other TFCs, and within the main modification for U2B transmission is the insertion
nominal set the most heavily loaded TFC should be the of dovetail interleaving that allows MUOS to exploit
one that delivers the specified RAB data rate. The 3GPP time diversity in a long coherence time UHF fading
downlink rate matching algorithm, with DTX on under- environment. The B2U has been modified in the way
loaded TFCs, is applied only to the nominal set. Thus, DPCCH bits are inserted into the waveform, and a
puncturing/repletion rates are set using the specified data modified rate matching algorithm efficiently
rate TFC (which would be the {4,0} TFC in our accommodates extended rate operation.
example). This protects the nominal set from excessive
extended rate puncturing.
The rate matching for each TFC in the exception set is REFERENCES
done in a fashion similar to the 3GPP uplink algorithm.
There is no DTX on exception set. For each exception [1] TS 25.211, "Physical channels and mapping of
set TFC, repetition and puncturing rates are set to transport channels onto physical channels
completely utilize all available DPDCH bits. Of course, (FDD)," 3GPP Technical Specification.
the exception set TFCs will have a diminished
performance relative to the nominal set. To overcome [2] TS 25.212, "Multiplexing and channel coding,"
this performance disadvantage, these TFC need a power 3GPP Technical Specification.
boost, again, much like the 3GPP uplink case. [3] TS 25.213. "Spreading and Modulation (FDD),"
Figure 3 illustrates the MUOS solution. A frame is 3GPP Technical Specification.
illustrated with T, P and D slots. As in 3GPP, the
DPCCH fields may be transmitted with inflated
amplitude relative to the data. All nominal set TFC
DPDCH symbols are transmitted with a fixed "nominal
set amplitude" and power is throttled back for under-
loaded TFCs (in the nominal set) via DTX bit insertion.
The nominal set TFC having the specified rate (the {4,0}
TFC achieving 32 kbps in our example) will have no
DTX, and will be transmitted at the nominal set
amplitude. An extended rate TFC is transmitted with a
large amplitude, in order to overcome its performance
loss.

6 of 6