EURASIP Journal on Applied Signal Processing ****:5, 772 782
c 2004 Hindawi Publishing Corporation
Multiple ARQ Processes for MIMO Systems
Haitao Zheng
Wireless Research Laboratory, Lucent Technologies, 791 Holmdel-Keyport Road, Holmdel, NJ 07733, USA
Email: abqnre@r.postjobfree.com
Angel Lozano
Wireless Research Laboratory, Lucent Technologies, 791 Holmdel-Keyport Road, Holmdel, NJ 07733, USA
Email: abqnre@r.postjobfree.com
Mohamed Haleem
Wireless Research Laboratory, Lucent Technologies, 791 Holmdel-Keyport Road, Holmdel, NJ 07733, USA
Email: abqnre@r.postjobfree.com
Received 4 December 2002; Revised 19 August 2003
We propose a new automatic repeat request (ARQ) scheme for MIMO systems with multiple transmit and receive antennas. The
substreams emitted from various transmit antennas encounter distinct propagation channels and thus have di erent error statis-
tics. When per-antenna encoders are used, separating ARQ processes among the substreams results in a throughput improvement.
Moreover, it facilitates the interference cancellation in certain MIMO techniques. Quantitative results from UMTS simulations
demonstrate that the proposed multiple ARQ structure yields more than 30% gain in link throughput.
Keywords and phrases: MIMO systems, automatic repeat request, throughput, wireless communication, UMTS.
tra c statistics, and quality-of-service requirements. Some
1. INTRODUCTION
of these adaptive techniques, relevant to this paper, are sum-
Third-generation cellular systems are being designed to sup-
marized below.
port high-speed packet data services. In the downlink, which
Multiple transmit and receive antennas. The use of mul-
has more stringent requirements in many of such services,
tiple antennas at each base station sector is already part of
high-speed packet access is provided through a shared chan-
every third-generation standard. In the downlink, speci -
nel where time-division multiplexing is used. Time slots are
cally, these antennas can be used to provide transmit diver-
assigned to users at speci c data rates through a scheduling
sity and/or to direct a beam towards the intended terminal.
algorithm based on the user data backlog and on channel
The deployment of multiple receive antennas at data ter-
quality indication (CQI) received via a feedback channel.1
minals is also being considered. The combination of mul-
Such a transmission scheme allows multiple users to share
tiple transmit and receive antennas will enable the imple-
the system resources e ciently by adapting to tra c and
mentation of a number of multiple-input multiple-output
channel variations and it also avoids possible resource lim-
(MIMO) techniques that promise spectacular increases in
itations that might occur if each user were allocated a ded-
throughput without the need for additional power or band-
icated code-multiplexed channel. Therefore, it has the po-
width [3, 4, 5].
tential to improve the capacity for delay-tolerant bursty ser-
Dynamic link adaptation through adaptive modulation
vices. Examples where this scheme will be implemented in-
and coding. Typically, each transmission in the downlink
clude the CDMA 1x EV-DO and 1x EV-DV and the UMTS
shared channel is at the maximum available power, with no
high-speed downlink packet access (HSDPA) [1, 2]. Several
power control. Therefore, link adaptation [6, 7], which ad-
advanced technologies are employed in high-speed downlink
justs the modulation and coding schemes (MCS), provides
transmission to improve link throughput or reduce packet
an e cient way of maximizing the instantaneous usage of
delay by adapting to the time-varying channel conditions,
the wireless channel. Speci cally, it enables the use of aggres-
sive MCSs when channel conditions are favorable while it re-
verts to MCSs that are more robust but with lower transmis-
1 Each terminal measures its channel condition and translates it into a
sion rates when channel conditions degrade. The base station
metric to be fed back to the serving base station.
Multiple ARQ Processes for MIMO Systems 773
selects the appropriate MCS based on the CQI for the user the overall throughput, while an overly conservative one fails
served at each time slot. We hereby refer to the MCS selec- to fully utilize the channel. In this case, the overall through-
tion process as the mapping design. put depends on the algorithms at both layers and only
cross-layer design can enable the most e cient use of the
Automatic repeat request (ARQ) or hybrid ARQ (HARQ).
The performance of MCS-based link adaptation largely de- channel.
pends on the accuracy of the CQI, which is di cult to main- In this paper, we address some of the key design issues as-
tain as velocity increases. The delay tolerance of many data sociated with the choice of the HARQ structure to be used for
services enables the use of retransmission schemes to re- MIMO physical layer transmission. We propose a new HARQ
cover erroneous packets. Recently, HARQ techniques have structure that matches the layered structure of the most pop-
been adopted by several wireless standardization bodies, for ular MIMO architectures [19]. Simulation results show that
example, 3GPP and 3GPP2. HARQ [8, 9, 10] can improve the performance sensitivity to the choice of HARQ depends
throughput performance, compensate for link adaptation er- on the aggressiveness of the transmissions and on the type of
rors, and provide a ner granularity in the rates e ectively CQI.
pushed through the channel. Upon detecting a transmission The paper is organized as follows. In Section 2, we de-
failure, mostly by cyclic redundancy check (CRC), the termi- scribe the layered architectures with per-antenna encoding.
nal sends a request to the base station for retransmission. The Modi cations to the conventional HARQ structure to t
delay due to packet acknowledgement can be signi cantly re- these layered architectures are discussed in Section 3. We
compare the performance of di erent HARQ structures in
duced by placing the HARQ functionality in the base station
(Node B in UMTS) rather than in the radio network con- Section 4. Conclusions are drawn in Section 5.
troller (RNC in UMTS). The packet decoder at the mobile
combines the soft information of the original transmission 2. LAYERED ARCHITECTURES WITH
with those of the subsequent retransmissions. The combined PER-ANTENNA ENCODING
signal has higher probability of successful decoding. In gen-
In order to approach the MIMO channel capacity in rich
eral, there are two ways of soft combining. With chase com-
multipath environments, the substreams radiated from the
bining, the base station repeatedly sends the same packet and
various transmit antennas should be uncorrelated [20, 21].
the receiver aggregates the energy from the (re)transmissions
Nonetheless, it may in practice be advantageous to jointly en-
to improve the signal-to-noise ratio (SNR) [11, 12]. A more
code them (Figure 1a). This has motivated a blossoming in-
sophisticated HARQ mechanism, named incremental redun-
terest in the design of space-time (vector) codes [22]. Clearly,
dancy (IR), transmits additional redundant information in
when the substreams are jointly encoded, they should share
each retransmission and gradually reduces the coding rate
a single CRC.
until successful decoding occurs [13, 14, 15]. Compared with
chase combining, IR requires larger receiver bu ers but it can The complexity of joint detection, however, explodes as
the number of transmit antennas grows large. As a result,
achieve better performance [16]. It also provides ner gran-
there has also been strong interest in devising alternative
ularity in the encoded rates and allows for better adaptation
approaches. One such approach is that of layered architec-
to channel variations.
tures, which incorporate multiple scalar encoders, one per
Scheduler. In a multiuser system where user channel con-
transmit antenna. In these architectures, input data is de-
ditions change over time, a scheduler can exploit those chan-
multiplexed into multiple substreams, which are then sep-
nel variations by giving certain priority to the users with
arately encoded and radiated from the various transmit an-
transitorily better channels. The scheduler critically impacts
tennas (Figure 1b). At the receiver, the substreams are succes-
the system performance. Several scheduling algorithms have
sively detected and cancelled [4, 5]. Speci cally, the informa-
been proposed in the literature to maximize the packet data
tion extracted from each substream is reencoded, interleaved,
throughput, subject to various fairness conditions [17].
and modulated to construct a replica of the transmitted sub-
The above technologies are tightly coupled. However,
since some of them reside in di erent layers, that is, HARQ stream. This replica, properly combined with the channel re-
sponse, is then subtracted from the overall received signal so
in the medium access control (MAC) layer and MIMO in
that if there are no errors the interference contribution
the physical layer, they are usually discussed and treated
of this substream is removed. The complexity of these archi-
separately. The evaluation of each technology fails to take
tectures increases more gracefully with the number of anten-
into account the performance improvement or degradation
nas. Furthermore, they can capitalize on existing scalar cod-
brought about by the other one. In particular, the link layer
ing formats.
performance of any MIMO algorithm is usually selected ac-
A layered architecture can approach the MIMO chan-
cording to the raw data rate at some operating point, for ex-
nel capacity if the data rates of the di erent transmit an-
ample, 10% packet error rate. However, when some level of
tennas are appropriately adjusted [23, 24]. This adjustment
channel uncertainty exists and the system supports HARQ, it
requires separate CQI, one per transmit antenna, and thus
may be bene cial to transmit aggressively at higher packet er-
the amount of feedback required increases linearly with
ror rates and recover channel errors through retransmissions
the number of transmit antennas. We hereby refer to it as
[18]. The throughput depends heavily on the transmission
per-antenna rate and CQI. Alternatively, a common CQI
strategy. An overly aggressive transmission could produce
and thus the same data rate can be used for all transmit
too many unsuccessful packet transmissions that diminish
774 EURASIP Journal on Applied Signal Processing
10
Input packet
Ergodic capacity (bps/Hz)
8
Channel coding &
interleaving 6
Substream 0 Substream 1 Substream 2
4
2
MIMO transmission
0
2 0 2 4 6 8
(a) MIMO with joint coding.
Average Es N0 (dB)
Per-antenna rate and CQI
Common rate and CQI
Input packet
Figure 2: Ergodic Shannon capacity with 4 transmit and 4 receive
antennas obtained via Monte Carlo simulation on a Rayleigh-faded
Multiplex
channel with no antenna correlation.
Substream 0 Substream 1 Substream 2
serve that in more than 70% of error events,2 only the sub-
Coding Coding Coding
streams from 1 or 2 transmit antennas are corrupted and thus
& & &
require a retransmission (Figure 4). However, upon an error
interleaving interleaving interleaving
event, an MSARQ receiver has to request a retransmission of
the entire packet because it relies on the single CRC over the
whole packet. Retransmitting substreams that have already
MIMO transmission
been correctly received wastes throughput. When multiple
per-antenna encoders are used, it becomes possible to re-
(b) MIMO with per-antenna coding.
move the constraint that the substreams radiated from mul-
tiple transmit antennas share a single ARQ process.
Figure 1: MIMO transmitter architecture with di erent coding For per-antenna MIMO encoding architectures, we
structures. herein propose to employ multiple ARQ processes, 1 for
each substream radiated from 1 transmit antenna or group
of antennas. This scheme is independent of the receiver-
processing algorithm and only requires that the receiver de-
antennas at the expense of some loss in capacity [23]. To codes substreams independently. We refer to this scheme as
illustrate this point, Figure 2 depicts the di erence between MIMO multiple ARQ (MMARQ). As shown in Figure 3b, a
the capacity with and without the constraint that the data CRC symbol is appended to each substream. At the receiver,
rate at each of the transmit antennas be equal, for the spe- each such substream is decoded and the associated CRC
ci c case of 4 transmit and 4 receive uncorrelated antennas is used to validate the content. Multiple acknowledgment
with Rayleigh fading. For the purpose of this paper, in any (NACK/ACK) indications are then sent back to the trans-
event, the most relevant feature of a layered architecture is mitter. After receiving these acknowledgements, the trans-
that it does not constraint the transmit antennas to be jointly mitter sends fresh packets from the transmit antennas that
encoded and share a unique CRC. have been successfully acknowledged and retransmits the
substreams that have been negatively acknowledged through
their associated transmit antennas. Hence, the HARQ opera-
3. HARQ MECHANISMS FOR MIMO SYSTEMS
tions at di erent transmit antennas are independent of each
If the MAC layer is unaware of the presence of MIMO at other. We focus on high-speed downlink data transmission
the physical layer, HARQ simply attaches a single CRC to so that the overhead due to multiple CRC symbols is neg-
the packet with such CRC encompassing the data radiated ligible. However, we need to consider the uplink signaling
from the various transmit antennas. We refer to this scheme, overhead due to multiple acknowledgements. For each ARQ
depicted in Figure 3a, as MIMO single ARQ (MSARQ). process, NACK/ACK requires an overhead of 1 bit plus error
Since substreams transmitted from di erent antennas en- protection redundancy. Therefore, the amount of ARQ feed-
counter distinct propagation channels, they have di erent er-
ror statistics. Using a typical channel propagation model with
4 transmit and 4 receive uncorrelated antennas [21], we ob- 2 An error event occurs when any of the substreams contains an error.
Multiple ARQ Processes for MIMO Systems 775
Input packet Input packet
Attach Multiplex
CRC
Attach Attach Attach
Multiplex CRC CRC CRC
Substream 0 Substream 1 Substream 2 Substream 0 Substream 1 Substream 2
Coding Coding Coding Coding Coding Coding
& & & & & &
interleaving interleaving interleaving interleaving interleaving interleaving
MIMO transmission MIMO transmission
(a) MSARQ transmitter. (b) MMARQ transmitter.
Figure 3: Transmitter structures of MSARQ and MMARQ.
0.8
describe the receiving procedures for both MMARQ and
MSARQ. The receiver decodes the transmitted substreams
sequentially following a certain order, which can be opti-
0.6
mized to achieve the best throughput performance. The rst
Event probability
substream is decoded from the overall aggregate received sig-
nal Y(t ). The information data S0 (t ), extracted from sub-
0.4 stream 0, is then reencoded, interleaved, and modulated to
construct a replica of the transmitted substream. This replica,
combined with the channel response, that is, F (S0 (t ), H(t )),
0.2 is then subtracted from Y(t ) so that the interference contri-
bution of this substream to the others is removed. This pro-
cedure is the so-called interference cancellation. The same
0 process is then applied to the remaining substreams, which
2 1 0 1 2 3 4 5 6 7 8
are thus successively extracted.
Average Es N0 (dB)
For MMARQ, the interference cancellation and HARQ
packet combining procedures can be blended advanta-
1 substream in error
2 substreams in error geously. In that case, the receiver would decode a substream
3 substreams in error
and use its associated CRC to validate the content. If this sub-
4 substreams in error
stream carries a retransmission packet and contains uncor-
rectable errors, the soft symbols of the packet would be com-
Figure 4: Probability distribution of the number of corrupted sub-
bined with those of the previous transmission(s) to extract
streams in an error event with 4 transmit and 4 receive uncorrelated
the information data. The receiver would then perform in-
antennas and frequency- at fading.
terference cancellation to remove the interference due to this
substream. Interference cancellation is performed regardless
of the results of the CRC validation; therefore, all the sub-
back overhead scales with the number of transmit antennas. sequent substreams can be decoded without waiting for the
When that number is large, grouping the transmit antennas retransmission of the current substream. However, the relia-
and assigning a single ARQ process to each group can reduce bility of the decoded data is much higher after HARQ packet
the signaling overhead. combining and, thus, using such data to reconstruct the sig-
Next, using per-antenna encoders with successive de- nal replicas for interference cancellation reduces error propa-
coding and cancellation at the receiver as an example, we gation. The detailed receiver procedure is shown in Figure 5.
776 EURASIP Journal on Applied Signal Processing
Received signal Y(t ), i = 0
Feedback
NACK(0 M 1) No i
Jiri Jan, Czech Republic
Keeping Your Copyright