Reliability and Low Latency for Cellular Wirelesssinghb1/URLLCmag.pdf · Abstract—The fifth generation (5G) cellular wireless technology is advancing towards enabling millisecond
Post on 14-Jul-2020
0 Views
Preview:
Transcript
Abstract—The fifth generation (5G) cellular wireless technology is advancing towards enabling millisecond level air-interface latency
with high reliability. This addresses the demand in many vertical sectors for machine-type of communication, enabling new use cases
and application areas, such as industrial automation and autonomous transportation. This paper gives a broad view of the status of this
development, including a description of the recent work in 3GPP Rel-15, and what the relevant next steps are. An overview of the main
technical enablers is also provided, including some new schemes for further performance enhancements. The novel proposals discussed
in the paper include downlink control information (DCI) bundling and repeated DCI bundling for blind transmissions to improve the
overall communication reliability of the control channel; analysis of different approaches with grant-free uplink; and a cooperative
relaying based approach for wireless motion control.
Index Terms—5G new radio (NR), ultra-reliable low latency communications (URLLC), automatic repeat request (ARQ), control
channels, grant-free transmissions, machine-type communications (MTC), verticals
INTRODUCTION
IFTH generation (5G) of wireless cellular systems brings significant improvement to communication between machines.
Originally, the goals of 5G research concerning machine communications was to enable high reliability, low latency
communications as well as the possibility to connect a very large number of devices with low energy consumption [1]. The latter
has been realized to a major extent already as part of the fourth generation, LTE, via the narrow band internet of things (NB-IoT)
technology and further enhancements are expected in future releases of 5G new radio (NR). Some aspects of the Ultra-Reliable
Low-Latency Communications (URLLC) are included already in the first 5G release of the Third Generation Partnership Project
(3GPP), but significant additional work remains in further stages of the evolution of 5G.
Earlier generations have steadily improved latency, achieving 10s of milliseconds (ms) in LTE, sufficient to serve the needs of
human to human communication. However, machines can in many cases significantly benefit from reliable, lower latency
communications. URLLC in the 5G system will thus enable emerging applications and services, such as wireless control and
automation in industrial environments, cooperative robot operations, communications for improved safety and efficiency in
transportation, and the tactile internet that allows controlling both real and virtual objects with real-time haptic feedback. This is
important for 5G, especially considering its potential application to verticals bringing new business to the whole telecommunication
industry.
This paper gives the overall picture on what the status of URLLC is and what to expect next. In the next section, we elaborate
on the need for URLLC via the example of motion control use case in a factory floor, as well as explain the status of standardization.
After that we expand on some of the enablers and outline related new enhancements for further evolution of URLLC. We then
conclude with a summary of the main points made in the paper.
WHERE ARE WE WITH URLLC?
URLLC has been identified as one of the three main usage scenarios for 5G. With URLLC, the key performance metrics are
low latency and high reliability. Generally, ultra-low latency refers to a few milliseconds or even lower, e.g., 1 ms end-to-end
latency between client and server on the user plane. With ultra-reliability, mostly we are talking about at least 99.999%, 99.9999%
or even higher successful packet delivery between device and server. In addition, network availability, referring to the downtime
of the network components such as base stations and core network elements, is also crucial for many applications. In this section,
The manuscript was submitted July xxth, 2018. This work was supported in part by the Business Finland in the context of the WIVE project.
Mikko A. Uusitalo and Zexian Li are with Nokia Bell Labs, Espoo, Finland (e-mail: mikko.uusitalo@nokia-bell-labs.com, zexian.li@nokia-bell-labs.com). Saeed R. Khosravirad is with Nokia Bell Labs, Canada (email: Saeed.Khosravirad@nokia-bell-labs.com). Harish Viswanathan is with Nokia Bell Labs, US
(email: Harish.Viswanathan@nokia-bell-labs.com).
Hamidreza Shariatmadari, Bikramjit Singh, Riku Jäntti, and Olav Tirkkonen are with Aalto University, Espoo, Finland (e-mail: hamidreza.shariatmadari@aalto.fi, bikramjit.singh@aalto.fi , riku.jantti@aalto.fi, olav.tirkkonen@aalto.fi)
Reliability and Low Latency for Cellular
Wireless
Mikko A. Uusitalo, Zexian Li, Saeed R. Khosravirad, Hamidreza Shariatmadari, Bikramjit Singh,
Riku Jäntti, Olav Tirkkonen and Harish Viswanathan
F
we will first discuss where such stringent requirements come from and then present the technology enablers available in the URLLC
tool box.
URLLC for motion control
As discussed in many publications, the main motivation of designing a system to meet these stringent requirements is expanding
the usage of the technology to vertical applications. Verticals include a very broad range of use cases from building automation,
factory of the future, e-Health, autonomous vehicles to electric-power distribution and so on. The connectivity requirements vary
substantially between different vertical applications. Below we will take motion control in the factory of the future as one example
to illustrate the machine operations and related requirements.
The main responsibility of a motion control system is controlling moving and/or rotating parts of machines in a well-defined
manner. Typically, a motion control system is composed of at least a motion controller, actuators and sensors. Within a motion
control system:
A motion controller sends desired set points to one or several actuators. The control commands are sent in a periodic
fashion.
The actuators take all the received set points and put them into an internal buffer.
At a well-defined time instance (usually referring as the “global sampling point” (GSP)), the actuators perform a
corresponding action (based on the latest set point in their buffer) on one or several processes (in this case usually a
movement or rotation of a certain component). At the same time, all sensors will send the latest data (e.g. measurement results) back to the motion controller.
It is quite typical that the sensors need to determine the current state of the processes at the same time instance as the actuators
taking actions. Therefore, the motion control puts a very high synchronicity requirement on all involved devices including sensors,
actuators, and motion controllers in terms of synchronicity which can be in the order of 1 us.
Based on the above example and other closed-loop control applications such as process automation and use of collaborative
robots the following communication requirements have been derived for 3GPP Rel-16 [2]:
Latency: from <0.5 ms to 10 ms.
Data rate: Low as messages are typically small (<50 bytes).
Reliability: Ultra high, ranging up to < 99.9999999%.
Positioning accuracy: Can be ultra-high, down to cm range.
Jitter: Variable though ultra-high, linked to isochronous operation and telegram delivery - from < 1 µs to 10 ms.
Traditionally, such industrial automation requirements are realized through cumbersome and expensive wired links e.g., using
fieldbus and Ethernet based solutions. Future industrial automation requires wireless connectivity solutions to enable highly
flexible and dynamic environment of production stations that can seamlessly be reconfigured based on production needs. However,
as further described below, the solutions standardized in 3GPP Rel-15 achieve a less stringent set of requirements sufficient for a
large number of use cases although not for the motion control example described above [3]. Additional enhancements are needed
which is expected to be standardized in Rel-16.
URLLC: 3GPP tool box
3GPP has been working on URLLC solutions since the beginning of Rel-15 New Radio (NR) work. Reliability is defined as the
success probability of transmitting a small packet within a certain latency. As the first step within radio access network (RAN)
group, the targeted URLLC reliability requirement for a data is 10-5 for 32 bytes with a user plane latency of 1 ms. The outcome
of Rel-15 URLLC work is summarized in the following table that captures the most important technology components that enable
achieving the stated requirements.
Table I. List of major URLLC technology components
Technology Component Short Description
Low latency New numerology, short
Transmission Time Interval
(TTI)/mini-slot
Short TTI size to ensure fast transmission of data packets.
New numerologies with flexible subcarrier spacing and
bandwidth leading to shorter slot; moreover mini-slot (1-13
OFDM symbols) based operation can further reduce latency.
Bi-directional slots for time
division duplex (TDD)
Bi-directional slots are a feasible solution to achieve low
latency communication in TDD where multiple switching
points can be configured within one slot. The time division
multiplexing of DL/UL control and data symbols in one slot
allows fast and energy efficient pipeline processing of control
and user data in the receiver.
Enhanced scheduling policy Non-slot based scheduling enables the minimum scheduling
unit to be on symbol level instead of slot level in conventional
scheduling algorithm and is an essential enabler to fulfill the
challenging latency targets in low frequency bands for TDD
and FDD NR. Providing better delay spread performance with a
small subcarrier spacing in wide area use cases, non-slot based
transmission is preferred over slot based transmission.
Pre-emption scheduling: When URLLC data arrives at the
gNB, it is immediately transmitted to the corresponding UE by
overwriting or pre-empting a part of an ongoing eMBB
transmission. eMBB transmission is recovered subsequently
through retransmissions. Dynamic multiplexing between
URLLC and non-URLLC traffic is preferred due to the
flexibility of resource usage leading to better spectral
efficiency.
Grant-free UL transmission Grant-free UL transmission scheme avoids regular handshake
delays between user equipment (UE) and gNB including at
least scheduling request and UL resource grant allocation, and
moreover relaxes the stringent requirements on the reliable
resource grant. Two types of UL grant-free transmission
schemes are supported in Rel-15. For the UL grant-free type 1,
UL data transmission without grant is based on RRC (re-
)configuration without any L1 signaling which is suitable for
deterministic URLLC traffic patterns. Traffic characteristics
can be well matched by appropriate resource configuration. The
UL grant-free type 2 allows additional L1 signaling for a fast
modification of semi-persistently allocated resources, which
enables flexibility of grant-free UL operation in terms of
URLLC traffic characteristics including but not limited to the
variable packet arrival time and/or packet size.
UE and gNB processing
time
Reduced UE and gNB processing time to ensure fast creation of
transport blocks for transmission as well as fast processing at
the receiver for fast feedback transmission by, for example,
placing dedicated pilot sequences at the beginning of the
scheduling resource.
High reliability Micro-diversity Single-user single-stream (i.e. Rank 1) transmission modes are
the most relevant for URLLC traffic. Clearly this is because our
goal is to improve performance on the lower tails of the SINR
distribution.
Macro-diversity: multi-
connectivity
Data duplication transmission from multiple cells to the same
UE to improve reliability through robustness against shadow
fading, blocking effects and cell failures. Packet data
convergence protocol (PDCP) layer duplications have been
specified in Rel-15.
Enhanced hybrid ARQ Blind repetition, a.k.a. K-repetition (where the same data packet
is transmitted K times without waiting for feedback), is part of
Rel-15. Time-diversity and frequency-diversity (in case of
frequency hopping) is generally enhanced by using HARQ with
We elaborate on some of the above features together with additional related enhancements that could be part of future releases.
URLLC TRANSMISSION SCHEMES
Radio access enhancements for enabling URLLC require careful attention to reliability of both data and control information.
The errors of data and control channels affect the overall communication reliability and latency. For example, unreliable
communication of feedback and downlink control information (DCI) can severely decrease reliability of ARQ-type transmissions
[1][4, [5]5]. For instance, errors in Channel Quality Indicator (CQI) may cause the transmission to use coding rate that is too high
resulting in increased block error rate. Overprovisioning of physical resources, e.g., repetition of control channel over physical
resources is a simple means to alleviate unreliability in communicating control information. However, such an approach can
significantly reduce the resource utilization efficiency and increase control channel blocking probability. The following is a list of
control information that are typically used in cellular wireless communication and need careful enhancements to enable URLLC:
Scheduling request (SR): A UE in the connected mode sends an SR to the gNB to be assigned with radio resources
over the data channel for uplink data transmissions. Missing the SR by the gNB increases the communication latency
as the UE needs to retransmit the SR.
Resource grant (RG): In scheduled-based transmissions, DCI including RG, sent by the gNB, informs a UE regarding
the assigned radio resources. In this regard, decoding the RG by the UE is prerequisite for utilizing the data channels.
ACK/NACK feedback: ARQ-type packet retransmissions are triggered upon receiving a NACK signal from the
receiver. However, the NACK signal might be erroneously decoded as ACK, which results in skipping the essential
data retransmissions and consequently reducing the communication reliability.
CSI report: Link adaption is employed to select an appropriate modulation and coding scheme (MCS) for data
transmissions. For this purpose, link adaptation needs to obtain CSI. In DL, this is achieved by receiving a CQI report
from the UE, which is derived by mapping the estimated SINR on the downlink to a CQI index according to a BLER
target. Wrong decoding of a CQI value as a higher one results in employing an MCS associated with a higher rate,
reducing the communication reliability.
Reliable communication of data and control information calls for employing enhanced data transmission and control signaling
protocols that offer efficient utilization of resources. In the following, we address the latency and reliability issues attributed to the
above control information and describe enhancements that apply to downlink and uplink URLLC.
Enhanced HARQ Operation
Reliability of HARQ protocol is tightly related to the reliability of the feedback channel. Specifically, probability of false positive
feedback detection (i.e., detecting NACK as ACK) at the transmitter node can result in packet outage at the MAC layer HARQ.
While the RLC layer ARQ is typically responsible for detecting such packet outages and to trigger a retransmission of the lost
packets, such RLC layer retransmission results in increased latency over RAN which is not tolerable by URLLC traffic.
Fig. 1 presents a perspective on the feedback detection strategies to overcome the unreliable feedback channel problem. It plots
the reliability of MAC layer HARQ operation with respect to feedback detection error rate and illustrates operating regions
associated with different feedback detection schemes. The plot is based on the assumptions that the maximum number of
soft combining. HARQ also adds robustness against potential
link adaption imperfections.
Enhanced control channel
reliability
Reliability requirements on data channel result in high
reliability requirements for control channel as well, for example
downlink control carrying resource grant, as well as uplink
control carrying feedback information such as ACK/NACK and
channel state information (CSI). The increased reliability can
be achieved for example by using much stronger coding, power
boosting and asymmetric signal detection techniques.
Interference mitigation Advanced UE interference mitigation receivers (e.g. MMSE-
IRC or NAICS). Network-based inter-cell coordinated
scheduling to reduce inter-cell interference.
transmission attempts allowed per packet is four and MCS selection for a packet targets 10% BLER. Region R1 reflects the
operation region for current HARQ operation similar to what is specified for the MAC layer of the 5G NR. Thus, ultra-reliable
HARQ operation with the current technology demands an ultra-reliable control channel to convey feedback bits to reduce false
feedback detection. That is traditionally provided using channel coding and repetition of the control channel over physical
resources.
The black dashed line depicts the HARQ reliability level achieved by four blind repetitions of each packet. Such a reliability
level can, for instance, be achieved by blind-HARQ that is specified in Error! Reference source not found.[6] to provide reliable
communication and further discussed in detail in the next section. Thus, area R5 reflects the reliability region that requires more
than the assumed maximum of four transmission attempts for the packet.
Regions R2 and R3 correspond to feedback detection schemes designed for reducing false ACK detection rate. Such schemes
are candidates for replacing the current HARQ operations when high reliability is required in presence of an unreliable feedback
channel. Particularly, the two feedback detection schemes that identify these two regions are as follows:
Backward composite feedback HARQ (BCF-HARQ): Conventional HARQ feedback is designed to acknowledge the
decoding success or failure of only the latest received packet. Alternatively, as proposed in [6][7], feedback report can
be a binary composite function of a given finite number of the most recent decoding acknowledgments from the same
HARQ process. For instance, by choosing the logical AND function as composite feedback function for ACKs, chances
for false positive detection of decoding status is decreased by orders of magnitude at the cost of slightly increasing the
average number of transmissions per data packet and the average experienced delay.
Asymmetric feedback detection (Asym-HARQ): Coherent detection of the feedback symbol can be adjusted to be more
biased towards detecting a NACK symbol so that wrongly detecting an ACK as a NACK is less likely [6, [6]8]. Such
asymmetric detection improves reliability at the cost of increasing likelihood of unnecessary retransmissions.
Area R2 refers to the operation region where BCF-HARQ is the desired approach for increased reliability. For minimal latency
experience, Asym-HARQ can alternatively be used in this operation region. In area R3, Asym-HARQ is the preferred feedback
based HARQ scheme. Area R4 refers to the blind spot of such feedback-based solutions. In such highly unreliable feedback channel
conditions, Asym-HARQ has no resource saving gain since the transmitter detects NACK at every feedback detection because of
the decision bias towards NACKs.
It is necessary to note here that in conditions with tight latency requirements, where feedback round trip time can become a
latency bottleneck, it is preferable to skip the feedback channel and operate with blind retransmission in all operation regions.
Figure 1 - MAC layer HARQ reliability regions for different feedback schemes.
Blind transmissions and DCI bundling
Blind transmissions are considered for 5G to enable performing data retransmissions without utilizing a feedback channel for
carrying ACK/NACK signals [9]. In this scheme, as shown in Figure 2(a), a transmitter sends 𝐾 replicas of a message using
Feedback detection error rate
R1R2
R3
90
99
99.9
99.99
99.999
Reliabilit
y o
f M
AC
HA
RQ
opera
tion [
%]
R4
R5
Formatted: English (United States)
different resource blocks (RBs), which are derived from symbols spanned over frequency and time domains, in order to exploit the
frequency and time diversity. Each data transmission is accompanied with a DCI to inform the receiver regarding the RBs carrying
the actual data and the employed MCS. The receiver can try to decode the received data if it successfully decodes the corresponding
DCI. The communication reliability can be improved by employing DCI bundling in case of sending data in consecutive time slots
using the same sizes of radio resources. The bundled DCI informs the receiver regarding the resource allocations for all the
consecutive data transmissions for a single message. This can be achieved by defining a frequency hopping pattern for data
transmissions and inclusion of few extra bits in the DCI as an indication of number of remaining transmissions (RT) for the
message, as shown in Figure 2(b). In this way, when a receiver successfully decodes a DCI, it can try to decode the message using
the data information received from all the remaining data transmission rounds. The performance of blind transmissions can be
further enhanced by employing the repeated DCI bundling. In this case, the initial DCI is repeated for 𝑚 times prior to the data
transmissions, as shown in Figure 2 (c). The UE can try to decode the message using all the data transmissions if it decodes at least
one of the repeated DCIs. This scheme also enables utilizing different number of DCI transmissions than the number of data
transmissions. For instance, fewer DCIs can be transmitted in case of having a reliable downlink control channel.
(a) (b) (c)
gNB UE
DCI(1, RT=3)
Data(1)
DCI(2, RT=2)
Data(2)
DCI(3, RT=1)
Data(3)
gNB UE
DCI(1)
Data(1)
DCI(2)
Data(2)
DCI(3)
Data(3)
gNB UE
DCI(1, RT=3)
Data(1)
Data(2)
Data(3)
DCI(1, RT=3)
DCI(1, RT=3)
Figure 2. Blind downlink transmission schemes using, (a) conventional DCI, (b) DCI bundling, and (c) repeated DCI bundling.
Figure 3 compares the overall communication reliabilities for the considered blind downlink transmission strategies. The
maximum of four transmission rounds is envisioned and each data transmission is self-decodable. The success probability of
decoding a message depends on the number of received data parts, i.e., 0.9, 0.994, 0.9999, and 0.999999, corresponding to receiving
one up to four data parts. It is evident that the DCI bundling and repeated DCI bundling can improve the communication reliability
for moderate reliability levels of the control channel.
Figure 3. Overall communication reliabilities for blind downlink transmission schemes.
Grant-Free Uplink Transmissions
To meet low latency constraint, Grant-Free (GF) transmission is a viable solution standardized in 3GPP Rel-15 where a UE
transmits its data instantly when it arrives in the buffer. The latency is reduced as the transmissions are done without sending an
SR (Scheduling Request), a BSR (Buffer Status Report) and subsequently receiving a RA (Resource Allocation) from the base
station. UL GF transmission is specified in 5G NR Release 15 [7]. The GF transmission occurs on a pre-defined resource pool and
is random, and therefore, collisions may happen. Hence, we propose multiple GF enhancements with an objective to increase GF
data transmission reliability.
The collisions can be minimized by means of multiple repeated GF transmissions, and is known as ‘K repetitions’ in 3GPP Release
15 [7]. UE transmits the same data packet K times, but possibly with different redundancy versions. Some repetitions may collide
and some may not. Thus, packet success probability increases. In [10][10], K repetitions is analyzed for different K values and
arrival rates. For sporadic transmissions, K repetitions helps to reduce collisions. However, collisions may increase for non-
sporadic UEs due to overpopulation of the resource.
If the transmissions are sporadic, then most of the transmissions even without repetitions will be collision-free. Only a few UEs
would suffer collisions, and to ensure their success, K repetitions can be utilized. This is done by implementing downlink feedback.
If the UE's transmission is successful, gNB communicates ACK. In case of erroneous transmission, and further if network is unable
to identify the UE, then gNB will not respond. If the UE does not receive feedback at the end of feedback waiting time, it proceeds
with blind K repetitions. If the network is able to decode UE’s identity for the erroneous packet, it sends NACK with RA for Grant-
Based (GB) access or just sending RA (implicit NACK). Hence, network is able to reduce unnecessary repetitions, which translates
to lower collisions.
With current RAN agreement Error! Reference source not found.[11], implicit NACK transmission has been agreed, but ACK
transmission is not specified for UL GF transmissions. In short, at the end of the transmission timer, UE assumes ACK if no NACK
is communicated by gNB to the UE. Such feedback constraint can reduce reliability. This is because in case of erroneous GF
transmission, gNB has no prior knowledge of UE’s GF transmission, and if gNB is unable to decode UE identity due to corrupted
transmission, NACK cannot be sent. As per current agreement, UE will assume transmission success for such failed transmissions
and reliability and/or latency will suffer. In addition, early termination of K repetitions, which also has been agreed cannot be
implemented without ACK feedback.
In principle, network can configure a UE with both GF and GB access at the same/neighboring time instants. The SR and GF
transmissions can occur in separate pools and in addition, GF packets can have pointers to its SR’s transmission resource so that
if GF packet is decoded, then SR will be ignored and vice versa. Further, K repetitions with and without feedback can be employed
for GF transmission, and similarly multiple SR transmissions can be utilized. GB offers collision-free access, but the bottleneck is
created by the transmission success of multiple rounds of communication - SR, RA, and data transmissions, and with no further
room for retransmissions in case of ultra-low latency constraint. On the other hand, multiple GF repetitions can be accommodated
in the same time-period as the limited GB transmission. Second, there is a single link dependency, i.e., on GF transmission itself.
The only drawback is that it suffers from random collisions. Figure 4Figure 4 depicts the Packer Error Rate (PER) for various GF
enhancements against different arrival rates. We see that each enhancement can outperform the others depending on the traffic,
channel conditions, reliability and latency constraint. However, each strategy possesses different implementation requirements
(e.g., feedback, multiple pools) which may inhibit the network for its utilization.
Figure 4. Packet Error Rate (PER) of four Grant-Free (GF) strategies are depicted against mean Poisson arrival per TTI. Latency
limit is 0.5 ms, and with 0.125 ms TTI size, maximum 4 TTIs are available. All packet lengths (GF data, GB data, SR, feedback)
are of a unit TTI. Processing times are excluded. Data and control channel transmissions are assumed with 10-3 and 10-4 BLER.
GF pool with five resource blocks is considered. In case of simultaneous access, first SR is transmitted in a pool with unit
resource block, and in next TTI, GF data is transmitted over the separate GF pool with four resource blocks. For strategies with
feedback triggered and simultaneous access, if there is no feedback, UE proceeds with GF transmission in the left over TTIs.
MESSAGE FORWARDING FOR REAL-TIME MOTION CONTROL
Time sensitive networks (TSN) or communication systems for industrial automation for closed loop control applications, such
as remote control of robotic manufacturing, and packaging and printing machines, face the most stringent reliability and latency
requirements, with packets delivered within 1 - 2ms at up to nine nines success probability. Specifically, as described earlier, in a
real-time motion control scenario, a controller node is connected through the wireless network to a set of UE’s, including actuator
and sensor nodes, for communicating small packets with sizes up to 50 Bytes per UE. Data traffic typically follows an isochronous
and periodic pattern where the controller sends actuation instructions for the UE’s, and then sensors will report the current state of
the control process to the controller.
In presence of strong channel impairments including small-scale and large-scale fading, extensive diversity is vital to fulfill the
ultra-high reliability requirement of TSN, while also meeting the stringent latency requirement. Fading channel coherence time
can typically span over several automation control cycles which precludes time diversity. Similarly, with limited bandwidth and
low delay spreads, frequency is normally not a reliable source of diversity. Therefore, other sources of diversity, such as spatial
diversity and multi-user cooperation should be exploited to reduce the required reliability-achieving SNR.
Multi-user diversity can be achieved in a periodic, fixed packet size transmission system by adapting the transmission spectral
efficiency of each packet, allocating more time frequency resources for devices with poor channel and fewer resources for devices
with better channel. As a result, the total transmission time to send all packets within a cycle is driven by the collective channel
state of all users. For such a system, channel sounding or feedback to learn the user SNRs is necessary and this itself can consume
significant resources, making it harder to send all packets within the cycle time. In [12][12], it is shown that the time overhead
from transmission of channel sounding pilot symbols can in fact reduce the reliability-achieving SNR, due to increasing the
diversity order.
For the case of strong shadowing, for example where an actuator robot is blocked by moving or static obstacles on the factory
floor, multi-hop transmission with cooperative relaying may be necessary to provide the blocked UE’s with reliable coverage
[13][13]. This calls for enabling cooperative relaying together with distributed multi-antenna transmission schemes such as cyclic
delay diversity (CDD) [14][14].
The above two approaches may be combined to create a robust wireless TSN system.
Figure 5. TDD mode cycle frame for motion control.
Figure 5Figure 5 illustrates an example TDD cycle-frame for such a system with N UE’s connected to a controller for
isochronous communication. Without loss of generality, we assume that the UE’s are indexed based on the decreasing order of
estimated SNR of their link to the controller BS. Both DL and UL frames include three phases as follows:
Phase 1: The first S UE’s (with strongest channels to BS, where S≤N) are scheduled with conservative MCS selection based
on the estimated channel, to guarantee successful transmission.
Phase 2: The messages for the remaining N-S UE’s are broadcasted. In case of DL, the messages can be concatenated
together prior to channel coding for better coding gain.
Phase 3: The broadcasted messages in phase 2 will be relayed in a distributed cooperative manner (e.g., using CDD) by the
UE’s who successfully decoded them earlier.
A common DCI is assumed which needs to be decoded by all the N UE’s. Switching the transmission direction between DL and
UL, as well as switching between broadcast and relaying phases, require a guard period to avoid interference among UE’s.
Simulations suggest that such a system can meet the TSN requirements.
CONCLUSIONS
For the past generations of mobile networks, cellular systems have mainly been optimized primarily for higher data rates and
capacity, and latency constraints have been driven only by human communication needs. However, by designing the system for
the more stringent URLLC requirements driven by machine communications, new opportunities emerge. The envisioned vertical
applications require revisiting the current protocols in cellular technologies for substantially improving the determinism of timely
service delivery. Rel-15 of 3GPP has made the initial steps towards reduced latency combined with higher reliability. This paper
presented an overview of the specified technology enablers, painting a picture of the status and expected future of 5G regarding
URLLC. Further enhancements in the coming releases required to meet more stringent requirements were also discussed.
As part of the URLLC technology evolution towards meeting more stringent reliability requirements, radio access protocol
design for data transmission and control signaling need to take more leaps forward for guaranteed packet delivery. From the view
point of reliability, this includes enhancements in HARQ operation and control signaling. It was shown that the DCI bundling and
repeated DCI bundling can improve the communication reliability for moderate reliability levels of the control channel. Analysis
of different approaches with grant-free uplink was provided and it was shown that hybrid grant free and grant based transmissions
are beneficial.
The requirements set for wireless industrial automation introduces the most stringent URLLC scenario. As pointed out in the
article, for case of wireless motion control, nine nines reliability figure is required for isochronous communication among a set of
sensor/actuator nodes and the controller entity. This can be realized with low latency only by exploiting large orders of macro
diversity, together with multi-user cooperation. We presented a framework for wireless motion control that considers careful
channel training to capture the multi-user diversity for the nodes with good quality link to the controller. Meanwhile, the weakest
Gua
rd p
erio
d
Gua
rd p
erio
d
Gua
rd p
erio
d
DL
ref
eren
ce
pilo
ts
DL
ref
eren
ce
pilo
ts
UL
ref
eren
ce
pilo
tsDistributed
relaying of
the DL
messages of
the N-S
weak nodes
DCIDL UE #1
DL UE #2
DL UE #S
DL
message of
the N-S
weak nodes
DC
I
DL phase 1 DL phase 2 DL phase 3
Gua
rd p
erio
d
UL
ref
eren
ce pilo
ts
UL UE #1
UL UE #2
UL UE #S
UL phase 1 UL phase 2 UL phase 3
UL
ref
eren
ce pilo
ts
UL UE #S+1
UL UE #S+2
UL UE #S+3
UL UE #NG
uard
per
iod
UL UE #S+1
UL UE #S+2
UL UE #S+3
UL UE #N
Motion control cycle frame
TD
D f
ram
e b
and
wid
th
users in the network are guaranteed a similar throughput through multi-hop cooperative forwarding of their message.
ACKNOWLEDGMENT
The authors would like to thank Business Finland for financial support in the context of the WIVE project.
REFERENCES
[1] A. Osseiran et al., “Scenarios for 5G Mobile and Wireless Communications: The Vision of the METIS Project,” IEEE Commun. Mag., vol. 52, no. 5, May
2014, pp. 26–35.
[2] 3GPP TR22.804, “Study on Communication for Automation in Vertical domains (Release 16)”, V16.0.0, June 2018.
[3] 3GPP TR38.913, “Study on Scenarios and Requirements for Next Generation Access Technologies”, V14.3.0, June 2017. [4] H. Shariatmadari et al., "Fifth-Generation Control Channel Design: Achieving Ultrareliable Low-Latency Communications," IEEE Vehic. Tech. Mag., vol.
13, no. 2, June 2018, pp. 84-93. [5] H. Shariatmadari et al., "Control channel enhancements for ultra-reliable low-latency communications," Proc. IEEE ICC Workshops, 2017.
[6] S. R. Khosravirad and H. Viswanathan, “Analysis of feedback error in Automatic Repeat reQuest,” arXiv preprint arXiv:1710.00649, 2017.
[7] 3GPP, “NR; Physical layer procedures for data”, TS 38.214 v15.1.0, Mar. 2018. [8] H. Shariatmadari et al., “Asymmetric ACK/NACK detection for Ultra-Reliable Low-Latency Communications,” Proc. IEEE EuCNC, 2018. [9] H. Shariatmadari et al.,” Statistical Analysis of Downlink Transmissions for Ultra-Reliable Low-Latency Communications,” Proc. IEEE ISWCS, 2018.
[10] B. Singh et al., "Contention-Based Access for Ultra-Reliable Low Latency Uplink Transmissions," in IEEE Wireless Commun. Lett., vol. 7, no. 2, Apr. 2018,
pp. 182-185.
[11] 3GPP RAN2#100 meeting, Chairman notes, page 142, http://www.3gpp.org/ftp/tsg_ran/WG2_RL2/TSGR2_100/Report/R2-1801701.zip, 2018. [12] R. Jurdi, S. R. Khosravirad, and H. Viswanathan, "Variable-rate ultra-reliable and low-latency communication for industrial automation," Proc. 2018 52nd
Annual Conference on Information Sciences and Systems (CISS), 2018.
[13] V. N. Swamy et al., “Cooperative communication for high-reliability low-latency wireless control,” Proc. IEEE Intl. Conf. on Commun. (ICC’15), June 2015.
[14] G. Bauch, "Capacity optimization of cyclic delay diversity," Proc. IEEE 60th Vehic. Techn. Conf., 2004.
BIOGRAPHIES
MIKKO A. UUSITALO [SM] (mikko.uusitalo@nokia-bell-labs.com) is head of the Research Department on Wireless Advanced
Technologies at Nokia Bell Labs Finland. He obtained an M.Sc. (Eng.) and a Dr.Tech. in 1993 and 1997, and a B.Sc. (economics)
in 2003, all from predecessors of Aalto University. He has been at Nokia since 2000 in various roles, including principal researcher
and head of International Cooperation at Nokia Research. He is a founding member of the CELTIC EUREKA and WWRF; the
latter, he chaired for 2004–2006.
ZEXIAN LI (zexian.li@nokia-bell-labs.com) is a senior specialist at Nokia Bell Labs, Finland. He received his B.S. and M.S.
degrees from Harbin Institute of Technology (HIT) and his Ph.D. degree from Beijing University of Posts and Telecommunications
(BUPT), P. R. China. From 2000 to 2005, he was a senior research scientist at the Centre for Wireless Communications, University
of Oulu, Finland. Since 2005, he has been with Nokia, Finland, focused on research and standardization activities on broadband
wireless communication systems, most recently on 5G and LTE-A Pro. He led the horizontal topic on direct D2D within the EU
FP7 METIS project and is leading the V2X radio interface design in the ongoing EU 5GCAR project. Currently his research
interests include 5G, wireless communications for vertical applications, and future wireless technologies for improving human life.
SAEED R. KHOSRAVIRAD (saeed.khosravirad@nokia-bell-labs.com) received his Ph.D. degree in telecommunications in 2015
from McGill University, Canada. Before that, he received a B.Sc. from University of Tehran, Iran, and a M.Sc. from Sharif
University of Technology, Iran, in 2007 and 2009 respectively. He is currently a Visiting Scholar at the Electrical and Computer
Engineering department of University of Toronto, Canada. He is with Nokia Bell Labs where his research fields of interest include
ultra-reliable wireless communication for industrial automation and radio resource management for 5G technology.
HAMIDREZA SHARIATMADARI (hamidreza.shariatmadari@aalto.fi) received the B.Sc degree in electrical engineering from the
University of Tabriz in 2009, and the M.Sc degree (with distinction) in communications engineering from Aalto University in
2013. He is currently pursuing a doctorate degree at Aalto University in the Department of Electrical Engineering. He is also
involved in a joint project between Aalto University and Nokia Bell Labs, aiming to enable ultra-reliable low-latency
communications in 5G networks. His research interests focus on development of wireless communication technologies for the
efficient support of machine-type communications.
BIKRAMJIT SINGH (Bikramjit.singh@aalto.fi) is a doctoral student at Aalto University, School of Electrical Engineering, Finland. He received his M.Sc. degree from Aalto University in Communication Engineering in 2014. His research interests include game-theoretic spectrum sharing, and resource management for URLLC and TSN systems.
RIKU JÄNTTI [M’02 - SM’07] (riku.jantti@aalto.fi) is an Associate Professor (tenured) in Communications Engineering and the
head of the department of Communications and Networking at Aalto University School of Electrical Engineering, Finland. He received his M.Sc (with distinction) in Electrical Engineering in 1997 and D.Sc (with distinction) in Automation and Systems Technology in 2001, both from Helsinki University of Technology (TKK). Prior to joining Aalto (formerly known as TKK) in August 2006, he was professor pro tem at the Department of Computer Science, University of Vaasa. Prof. Jäntti is a senior member of IEEE and associate editor of IEEE Transactions on Vehicular Technology. He is also IEEE VTS Distinguished Lecturer (Class 2016). The research interests of Prof. Jäntti include radio resource control and optimization for machine type communications, Cloud based Radio Access Networks, spectrum and co-existence management and quantum communications.
HARISH VISWANATHAN is Head of the Radio Systems Research Group at Nokia Bell Labs. He leads an international team of
researchers investigating various aspects of wireless communication systems, and in particular, 5G. In his prior role, as a CTO
Partner he was responsible for advising the Corporate CTO on Technology Strategy through in-depth analysis of emerging
technology and market needs. He joined Bell Labs in 1997 and has worked on multiple antenna technology for cellular wireless
networks, mobile network architecture, and M2M. He received the B. Tech. degree from the Department of Electrical Engineering,
Indian Institute of Technology, Madras, India and the M.S. and Ph.D. degrees from the School of Electrical Engineering, Cornell
University, Ithaca, NY. He holds more than 50 patents and has published more than 100 papers. He is a Fellow of the IEEE and a
Bell Labs Fellow.
top related