A Novel Uplink Data Transmission Scheme For
Small Packets In Massive MIMO System
Ronggui Xie
1
, Huarui Yin
1
, Zhengdao Wang
2
, and Xiaohui Chen
1
1
Department of Electronic Engineering and Information Science, University of Science and Technology of China
2
Department of Electrical and Computer Engineering, Iowa State University
Abstract—Intelligent terminals often produce a large num-
ber of data packets of small lengths. For these packets, it
is inefficient to follow the conventional medium access control
(MAC) protocols because they lead to poor utilization of service
resources. We propose a novel multiple access scheme that targets
massive multiple-input multiple-output (MIMO) systems based
on compressive sensing (CS). We employ block precoding in
the time domain to enable the simultaneous transmissions of
many users, which could be even more than the number of
receive antennas at the base station. We develop a block-sparse
system model and adopt the block orthogonal matching pursuit
(BOMP) algorithm to recover the transmitted signals. Conditions
for data recovery guarantees are identified and numerical results
demonstrate that our scheme is efficient for uplink small packet
transmission.
I. INTRODUCTION
As intelligent terminals such as smart phones and tablets
get more popular, they produce an increasing number of data
packets of short lengths. Modern mobile applications that
produce such small packets include instant messaging, social
networking, and other services [1], [2]. Although the lengths
of messages are relatively short, small packet services put
great burden on the communication network. Two kinds of
messages contribute to the traffic of small packets: one is the
small packets of conversation produced by active users that
occupy only a small percentage of the total online users [2];
the other is the signaling overheads needed to transmit these
conversation packets [3].
In current wireless communication systems, a user follows
the medium access control (MAC) protocols to obtain the
service resources. Either resources are preallocated to the users
in a noncompetitive fashion, or certain random access scheme
with collision resolution is used. For small and random pack-
ets, the reservation-based approach is inefficient in resource
utilization due to irregularity of the packets. The collision-
resolution based approaches, on the other hand, can suffer
from too many retransmissions due to frequent collisions.
Recently, massive multiple-input multiple-output (MIMO)
was studied as a way to improve the system throughput of
cellular systems [5]- [8]. In massive MIMO systems, the num-
ber of antennas at the base station (BS) can be more than the
number of active single-antenna users that are simultaneously
served. When the number of antennas at BS is large, the
different propagation links from the users to the BS tend to be
orthogonal, and the large amount of spatial degrees of freedom
are useful for mitigating the effect of fast fading [6], [7].
Overall, massive MIMO technique provides higher data rate,
better spectral and energy efficiencies [8]. All these advantages
make massive MIMO a promising technique.
In this paper, we propose a novel uplink small packet trans-
mission scheme based on precoding at the transmitters and
sparsity-aware detection at the receiver. The main motivation is
to allow for a large number of users to transmit simultaneously,
although each user may be transmitting only a small amount
of data. Besides frame-level synchronization, no competition
for resources or other coordinations are required. This saves
the signaling overhead for collision resolution, and improves
the resource utilization efficiency.
The contributions of our work are as follows:
1) block-sparse system model is established: We apply
block precoding at each transmitter in time domain, and
by considering the user activities, develop a block-sparse
system model [9]- [11], taking full advantages of the
additional structure property of the signals to recover.
2) conditions for signal recovery are given: The result
of our analyses about the block orthogonal matching
pursuit (BOMP) algorithm is milder than those in the
related work in [10]. Furthermore, we characterize the
data recovery condition from information theoretic point
of view.
Thanks to the precoding operation and our sparsity-aware
detection algorithms, our scheme enable the system to support
more active users to be simultaneously served. The number of
active users can be even more than the number of antennas at
BS. This is of great practical significance for networks offering
small packet services to a large number of users.
Applications of compressive sensing (CS) to random MAC
channels have been considered in [12]- [15]. In [12], CS based
decoding scheme at the BS has been used for the multiuser
detection task in asynchronous random access channels. A
technique based on CS for meter reading in smart grid is
proposed in [13], and its consideration is limited to single-
antenna systems. Besides, a novel neighbor discovery method
in wireless networks with Reed-Muller Codes has been pro-
posed in [15], where CS technique is also adopted. All the
referred works depend on the idea that the MAC channel is
sparse, and all their works are classified to initial category of
CS, where no structure property have been taken into account.