05. Guidance, Navigation and Control

Download PDF: GNC_Jan2019.pdf

Introduction

The Guidance, Navigation & Control (GNC) subsystem includes both the components used for position determination and the components used by the Attitude Determination and Control System (ADCS).

In Earth orbit, onboard position determination can be provided by a GPS receiver. Alternatively, ground based radar tracking systems can also be used. If onboard knowledge is required, then these radar observations can be uploaded and paired with a suitable propagator. Commonly, the USAF publishes two line element (TLE) sets1  which are paired with a SGP4 propagator2. In deep space, position determination is performed using the Deep Space Network (DSN) and an onboard radio transponder 3.

AADCS includes sensors used to determine attitude and attitude rate, such as star trackers, sun sensors, horizon sensors, magnetometers, and gyros. Actuators are designed to change a spacecraft’s attitude. Common spacecraft actuators include magnetorquers, reaction wheels, and thrusters. There are many attitude determination and control architectures and algorithms suitable for use in small spacecraft4.

The continuing trend in small spacecraft GNC is the miniaturization of existing technologies. While 3-axis stabilized, GPS-equipped, 100 kg class spacecraft have been flown for decades, it has only been in the past few years that such technologies have become available for micro- and nano-class spacecraft. Table 5-1 summarizes the current state-of-the-art of performance for GNC subsystems in small spacecraft. Performance greatly depends on the size of the spacecraft and values will range for nano- to micro-class spacecraft.

The author would like to highlight that the presented tables are not intended to be exhaustive but to provide an overview of current state-of-the-art technologies and their development status for this small spacecraft subsystem. There is no intention of mentioning certain companies and omitting others based on their technologies.

Table 5-1: The state of the art for GNC subsystems
Component Performance TRL Status
Reaction Wheels 0.001 – 0.3 Nm peak torque, 0.015 – 8 Nms storage 9
Magnetorquers 0.1 Nm peak torque, 1.5 Nms storage 9
Star Trackers 25 arcsec pointing knowledge 9
Sun Sensors 0.1° accuracy 9
Earth Sensors 0.25° accuracy 9
Gyroscopes 1°h-1 bias stability, 0.1°h-1/2 random walk 9
GPS Receivers 1.5 m position accuracy 9
Integrated Units 0.002° pointing capability 9

State of the Art

Integrated Units

bct_xact Integrated ADCS Unit
BCT XACT Integrated ADCS Unit. Image Courtesy of Blue Canyon Technologies.

Integrated units combine multiple different attitude and navigation components into a single part to provide a simple, single-component solution to a spacecraft’s GNC requirements. Typical components included are reaction wheels, magnetometers, magnetorquers, and star trackers. The units often include built-in attitude determination and momentum management algorithms. Table 5-2 describes some of the integrated units currently available. Blue Canyon Technologies’ XACT is currently flying on the NASA-led missions MarCO and ASTERIA, both of which are 6U platforms, and have also flown on 3U missions (MinXSS was deployed from NanoRacks in February, 2016).

Table 5-2: Integrated GNC Units
Product Manufacturer Mass (kg) Components Pointing Accuracy Status
High-Precision Attitude Determination and Control System AAC-Clyde 0.086 ADCS processor, sensors and actuators, magnetorquers, GPS receiver chip, single and 3-axis reaction wheels, µPPT 0.5° 9
Inertial Reference Module (IRM) Tyvak 0.61 2 Orthogonal Star Trackers, 3-Axis MEMS Gyro, Reaction Wheels (x3), Torque Coils (x3), C&DH processor, ADCS processor 0.057° 1-σ 9
MAI-400 Adcole Maryland Aerospace 0.694 3 reaction wheels, 3-axis magnetometer, 2 IREHSs, 3 torque rods 9
MAI-401 Adcole Maryland Aerospace 0.56 3 reaction wheels, 3-axis magnetometer, star tracker, 3 torque rods <0.1° 7
MAI-500 Adcole Maryland Aerospace 0.694 3 reaction wheels, 3-axis magnetometer, 2 star trackers, 3 torque rods <0.1° 7
XACT Blue Canyon Technologies 0.91 3 reaction wheels, 3-axis magnetometer, star tracker, 3 torque rods 0.007° 9
XACT-50 Blue Canyon Technologies 1.23 3 reaction wheels, 3-axis magnetometer, star tracker, 3 torque  rods 0.007° 9
iADCS-100 Berlin Space Technologies 0.345 Star tracker, 3 gyro modules, 3 reaction wheels, 3 magnetorquers, optional sensors <<1° 9

Reaction Wheels

sinclair_rw003 Reaction Wheel
Sinclair Interplanetary RW-0.03 Reaction Wheel. Image Courtesy of Sinclair Interplanetary.

Miniaturized reaction wheels provide small spacecraft with precision pointing capability. Reaction wheels can provide arbitrary torques limited by the wheel’s peak torque, momentum capacity, and wheel dead-band. Table 5.3 lists a selection of high-heritage miniature reaction wheels, and the Figure here depicts one of the wheels offered by Sinclair Interplanetary. With the exception of three units, all of the reaction wheels listed in Table 5-3 have spaceflight heritage. For example, Blue Canyon’s RWp500 has been flying on NASA’s CYGNSS mission since 2015, and Millennium Space Systems has 20 RWA1000s in orbit. For full three-axis control, a spacecraft requires three wheels. However, a four wheel configuration is often used to provide fault tolerance5.  Due to parasitic external torques, reaction wheels need to be periodically desaturated using an actuator that provides an external torque, such as thrusters or magnetorquers 6.

Table 5-3: Reaction Wheels
Product Manufacturer Mass (kg) Peak Torque (mNm) Momentum Capacity (Nms) Radiation Tolerance (krad) TRL Status
10SP-M Surrey Satellite Technology 0.96 11 0.42 5 9
100SP-O Surrey Satellite Technology 2.6 110 1.5 5 9
RW-0.03 Sinclair Interplanetary 0.185 0.5 0.04 20 9
RW-0.003 Sinclair Interplanetary <0.05 1 0.005 10 6
RW-0.01 Sinclair Interplanetary 0.12 1 0.018 20 9
RW3-0.06 Sinclair Interplanetary 0.226 20 0.18 20 9
MAI-400 Reaction Wheel Adcole Maryland Aerospace 0.11 0.635 .0111 Unkn. 9
MicroWheel Blue Canyon Technologies 0.13 4 0.015 Unkn. 9
RWp500 Blue Canyon Technologies 0.75 25 0.5 Unkn. 9
RWp050 Blue Canyon Technologies 0.24 7 0.05 Unkn. 6
RWp100 Blue Canyon Technologies 0.35 7 0.1 Unkn. 6
SmallSat Reaction Wheel ClydeSpace 1.5 40 Unkn. 10 9
RWA1000 Millenium Space Systems Unkn. 1000 0.1 Unkn. 9

Magnetorquers

ZARM Technik Magnetorquers for Micro-Satellites
ZARM Technik Magnetorquers for Micro-Satellites. Image Courtesy of ZARM.

Magnetorquers are an established technology used in small spacecraft and can provide control torques perpendicular to the local external magnetic field. Table 5-4 lists a selection of high heritage magnetorquers and the Figure shown illustrates some of ZARM Technik’s product offerings. Magnetorquers are often used in combination with wheels to remove excess momentum. As control torques can only be provided in the plane perpendicular to the local magnetic field, full 3-axis stabilization is not possible at any given time.

Table 5-4: Magnetorquers
Product Manufacturer Mass (kg) Peak Dipole (A m2) Radiation Tolerance (krad) TRL Status
MTR-5 Surrey Satellite Technology 0.5 5 5 9
MT0.1-1 ZARM 0.003 0.1 Unkn. 9
MT1-1 ZARM 0.060 1 Unkn. 9
0-1-1 Spaceflight Industries 0.727 15 Unkn. 9
Electromagnet (Type A) Adcole Maryland Aerospace 0.018 0.15 Unkn. 9
TQ-40 Sinclair Interplanetary 0.825 48 Unkn. 9
TQ-15 Sinclair Interplanetary 0.4 19 Unkn. 9
SatBus MTQ NanoAvionics <0.2 0.2 Unkn. 9

Thrusters

Thrusters used for attitude control are described in Propulsion Chapter. Pointing accuracy is determined by minimum impulse bit, and control authority by thruster force.

Star Trackers

sstl_procyon Star Tracker
SSTL Procyon Star Tracker. Image Courtesy Surrey Satellite Technology Ltd.

A star tracker can provide an accurate, standalone estimate of the spacecraft’s attitude by comparing a digital image captured with a CCD or CMOS sensor to an onboard star catalog7. Table 5.5 lists some models suitable for use on a small spacecraft, one of which is depicted in Figure 5.4. Sinclair Interplanetary has flown about 38 ST-16RT2 units, Blue Canyon Technologies has flown the Extended NST onboard the DARPA High Frequency Receiver Experiment, and many other star trackers also have notable flight heritage.

Table 5-5 Star Trackers
Product Manufacturer Mass  incl. baffle (kg) Accuracy (arcsec) Radiation Tolerance (krad) TRL Status
Rigel-L Surrey Satellite Technology 2.2 25 5 9
Procyon Surrey Satellite Technology 1.7 30 5 9
ST-16 Sinclair Interplanetary 0.12 74 9 9
ST-16RT2 Sinclair Interplanetary 0.185 55 Unkn. 9
MAI-SS Space Sextant Adcole Maryland Aerospace 0.282 27 75 9
Standard NST Blue Canyon Technologies 0.35 40 Unkn. 9
Extended NST Blue Canyon Technologies 1.3 40 Unkn. 9
ST200 Berlin Space Technologies 0.04 10 11 9

Magnetometers

NSS Magnetometer
NSS Magnetometer. Image Courtesy of NewSpace Systems.

Magnetometers provide a measurement of the local magnetic field, and this measurement can be used to provide both estimates of attitude8 and also orbital position. The vast majority of CubeSats use COTS magnetometers and improve their performance with software Table 5-6 provides a summary of some 3-axis magnetometers available for small spacecraft.

Table 5-6: Magnetometers
Product Manufacturer Mass (kg) Resolution (nT) Orthogonality (°) Radiation Tolerance (krad) TRL Status
Magnetometer New Space Systems 0.085 10 <1 10 9
Magnetometer Surrey Satellite Technology 0.14 10 1 10 (Si) 9
3-axis Magnetometer Adcole Maryland Aerospace Unkn. Unkn. Unkn. Unkn. 9
MAG-3 SpaceQuest 0.1 Unkn. <1 10 9
MicroMag3 PNI Corp 0.2 15 <1 Unkn. 9
MAG-3 Three-Axis Magnetometer SpaceQuest 0.1 Unkn. <1 10 9

Sun Sensors

Adcole Coarse Sun Sensor Detector (Cosine Type)
Adcole Coarse Sun Sensor Detector (Cosine Type). Image Courtesy of Adcole Corporation.

Sun sensors are used to provide an estimate of the location of the Sun in the spacecraft body frame, which in turn can be used to estimate attitude. A digital two-axis sun sensor can provide perfectly fine sun vector solutions, but multiple sensors are typically used in case a spacecraft is “lost in space.” Fine sun sensors provide a full 2-axis estimate of Sun location9  and a spacecraft would require a minimum of six and a minimum of four are required.

Table5-7: Sun Sensors
Product Manufacturer Mass (kg) Accuracy (°) Radiation Tolerance (krad) TRL Status
Fine (digital) Sun Sensor New Space Systems 0.035 0.1 10 9
Analog Sun Detector Adcole Maryland Aerospace 0.068 0.75 Unkn. 9
CSS-01 Space Micro 0.0141 5 Unkn. 9
BiSon64 Lens Research & Development 0.0217 0.5 1100 8
BiSon64-B Lens Research & Development 0.0217 0.5 1100 8
BiSon74-ET-RH Lens Research & Development 0.0245 0.7 1100 ~6
SS-411 Sinclair Interplanetary 0.034 0.1 20 9
DSS1 NanoAvionics 0.015 0.5 Unkn. 9

Horizon Sensors

MAI-SES.
MAI-SES. Image Courtesy of Maryland Aerospace Inc.

Horizon sensors can be simple infrared horizon crossing indicators (HCI) or more advanced thermopile sensors can be used to detect the temperature differences between the poles and the equator. For terrestrial applications, these sensors are referred to as Earth Sensors, but can be used for other planets. Examples of such technologies are described in Table 5-8.

Table 5-8: Horizon Sensors
Product Manufacturer Mass (kg) Accuracy (°) TRL Status
MAI-SES Static Earth Sensor Adcole Maryland Aerospace 0.033 0.25 9
Mini Digital HCI Servo 0.050 0.75 9

Gyros

LN-200S Fiber Optic Gyro and IMU.
LN-200S Fiber Optic Gyro and IMU. Image Courtesy Northrop Grumman Corporation.

GGyroscopes provide a measurement of angular velocity. The main gyro types used in small spacecraft are fiber optic gyros (FOGs) and MEMS gyros, with FOGs offering better performance at a mass and cost penalty10. Table 5.9 lists a sample of gyros available for small spacecraft.

Table 5-9: Gyros
Product Manufacturer Type Mass (kg) Bias Stability (°h-1) Random Walk (°h-1/2) Radiation Tolerance (krad) TRL Status
MIRAS-01 Surrey Satellite Technology 3-axis MEMS 2.8 10 0.6 5 9
LN-200S Northrop Grumman 3- axis FOG 0.75 1 0.07 10 9
ADIS16405 Analog Devices 3-axis MEMS 0.016 25 2.0 Unkn. 9
MASIMU04 Micro Aerospace Solutions 3-axis MEMS 0.03 0.6 Unkn. Unkn. Unkn.

GPS Receivers

 NovaTel OEM615 Dual-Frequency GNSS Receiver.
Figure 5.9: NovaTel OEM615 Dual-Frequency GNSS Receiver. Image Courtesy of NovAtel Inc.

For LEO spacecraft, GPS receivers are now the primary method for performing orbit determination, replacing ground based tracking methods. Onboard GPS receivers are now considered a mature technology for small spacecraft, and some examples are described in Table 5.10. There is a new generation of chip-size COTS GPS solutions, for example the NovaTel OEM 719 board has replaced the ubiquitous OEMV1.

GPS accuracy is limited by propagation variance through the exosphere and the underlying precision of the civilian use C/A code11.  GPS units are controlled under the Export Administration Regulations (EAR) and must be licensed to remove COCOM limits12.

Table 5-10: GPS Receivers
Product Manufacturer Mass (kg) Accuracy (m) Radiation Tolerance (krad) TRL Status
SGR-05U Surrey Satellite Technology 0.04 10 5 9
SGR-10 Surrey Satellite Technology 0.95 10 10 9
OEM615 Novatel 0.021 1.5 Unkn. 9
piNAV-NG SkyFox Labs 0.024 10 Unkn. 9

Deep Space Navigation

General Dynamics SDST
Figure 5.10: General Dynamics SDST. Image Courtesy of General Dynamics.

In deep space, navigation is performed using radio transponders in conjunction with the Deep Space Network (DSN). As of 2018, the only deep space transponder with flight heritage that is suitable for small spacecraft is the JPL-designed and General Dynamics-manufactured Small Deep Space Transponder (SDST). JPL has also designed IRIS V2, which is a deep space transponder that is more suitable for the CubeSat form factor. Table 5.11 details these two radios, and the SDST is illustrated in Figure 5.8. IRIS V2, derived from the Low Mass Radio Science Transponder (LMRST), is currently flying on the MarCO CubeSats and is scheduled to fly on INSPIRE 13.

Table 5-11: Deep Space Transponders
Product Manufacturer Mass (kg) Bands TRL Status
SDST General Dynamics 3.2 X, Ka 9
IRIS V2 JPL 0.4 X, Ka, S, UHF 9

Atomic Clocks

AAtomic clocks have been used on larger spacecraft in LEO for several years now, however integrating them on small spacecraft is relatively new. The conventional method for spacecraft navigation is a two-way tracking system of ground-based antennas and atomic clocks. The time difference from a ground station sending a signal and the spacecraft receiving the response can be used to determine the spacecraft’s location, velocity, and path. This is not a very efficient process, as the spacecraft must wait for navigation commands from the ground station instead of making real time decisions, and the ground station can only track one spacecraft at a time, as it must wait for the spacecraft to return a signal 14. In deep space navigation, the distances are much greater from the ground station to spacecraft, and the accuracy of the radio signals needs to be measured within a few nanoseconds.

JPL’s Deep Space Atomic Clock (DSAC) project plans to launch a prototype of a small, low mass (16 kg) atomic clock based on mercury-ion trap technology, which underwent demonstration testing in the fall of 2017. The project aims to produce a <10 kg configuration in the second generation. The DSAC is slated for will launch in 2019 as a hosted payload on General Atomic’s Orbital Test Bed spacecraft aboard the U.S. Air Force Space Technology Program (STP-2) mission 15.

More designers of small spacecraft technology are developing their own version of atomic clocks and oscillators to be used in space and need to ensure they are properly synchronized. They are designed to fit small spacecraft, missions that are power and volume limited, and those that require multiple radios. Table 5-12 lists the atomic clocks and oscillators available for small spacecraft missions.

Iris series 1x1 OCXO for LEO
Iris series 1×1 OCXO for LEO. Image Courtesy of Bliley Technologies.

Bliley Technologies has developed a miniature Half-DIP package low power Oven-Controlled Crystal Oscillation (OCXO) and an Iris series 1″x1″ OCXO for LEO (Figure 5.9) that is desirable for power constrained missions. The Half-DIP package has 135 mW power consumption, and superior close in phase noise of -125 dBc/Hz at 10 Hz 16. This part is characterized at TRL 6, however the components have not been radiation tested. The Iris series can range from 10-100 MHz in frequency and has a stability vs temperature performance of +/-25ppb with a sine output and a radiation tolerance of 38 kRAD TID.

Table 5‑12: Atomic Clocks
Product Manufacturer Dimensions (mm) Power Consumption Frequency Range (MHz) TRL Status
Miniature Half-DIP package Low Power OCXO Bliley Technologies, Inc. Up to 12 x 12 x 10 135 – 180 mW at steady state 10 – 60 6
Iris Series 1″x1″ OCXO for LEO Bliley Technologies, Inc. 19 x 11 x 19 1.5 W at steady state 10 – 100 6
Ultra Stable Oscillator AccuBeat, Ltd. 120x120x120 3.8 W Unkn. 6
9635QT Microsemi 33 x 33 x 33 Unkn. Unkn. 6
Miniature Atomic Clock (MAC) SA.3Xm Microsemi 50.8 x 50.8 x 18 5 – 8 W 10 Unkn.
Space Chip Scale Atomic Clock (CSAC) Microsemi 40 x 35 x 11 <120 mW 10 9

On the Horizon

Technological progress in the area of guidance, navigation, and control is slow. Given the high maturity of existing GNC components, future developments in GNC are mostly focused on incremental or evolutionary improvements, such as decreases in mass and power, increases in longevity and/or accuracy. This is especially true for GNC components designed for deep space missions, where small spacecraft-focused missions have only very recently been proposed. However, in a collaborative effort between the Swiss Federal Institute of Technology and Celeroton, there is progress being made on a high-speed magnetically levitated reaction wheel for small satellites (Figure 5.10). The idea is to eliminate mechanical wear and stiction by using magnetic bearings rather than ball bearings. The reaction wheel implements a dual hetero/homopolar, slotless, self-bearing, permanent-magnet synchronous motor (PMSM). The fully active, Lorentz-type magnetic bearing consists of a heteropolar self-bearing motor that applies motor torque and radial forces on one side of the rotor’s axis, and a homopolar machine that exerts axial and radial forces to allow active control of all six degrees of freedom. It is capable of storing 0.01 Nm of momentum at a maximum 30,000 rpm, applying a maximum torque of 0.01 Nm 17 .

Summary

Small spacecraft GNC is a mature area, with many previously flown, high TRL components offered by several different vendors. Progress in developing integrated units will offer simple, single vendor, modular devices for ADCS which will simplify GNC subsystem design. Other areas of GNC have potential for improvements as more research is being conducted. For example, a team at the University of Michigan is developing a multi-algorithmic hybrid ADCS system for CubeSats that can implement multiple estimation and control algorithms 18. Another team from Johns Hopkins University is conducting ground simulations of docking, charging, relative navigation, and deorbiting for a fully robotic CubeSat19 . The RANGE mission from Georgia Institute of Technology is a pair of 1.5U CubeSats that will improve the relative and absolute positioning capabilities of nanosatellites 20.

The rising popularity of smallsats in general, and CubeSats in particular, means there is a high demand for components, and engineers are often faced with prohibitive prices. The Space Systems Design Studio at Cornell University is tackling this issue for GNC with their PAN nanosatellites. A paper by Choueiri, et al. outlines an inexpensive and easy-to-assemble solution for keeping the ADC system below $2,500 21Lowering the cost of components holds exciting implications for the future, and will likely lead to a burgeoning of the smallsat industry.

For Feedback solicitation, please email: arc-sst-soa@mail.nasa.gov. Please include a business email so someone may contact you further.6

1.
Shepherd LCG and AFSC. Space Surveillance Network. Presented at the: Shared Space Situational Awareness Conference, Colorado Springs, CO; 2006.
2.
Vallado DA, Crawford P, Hujsak R, Kelso TS. Revisiting spacetrack report 3. 2006;6753:2006-2006.
3.
Thornton CL, Border JS. Radiometric Tracking Techniques for Deep-Space Navigation. John Wiley & Sons; 2003.
4.
Wertz JR. Spacecraft Attitude Determination and Control. Vol 73. Springer Science & Business Media; 2012.
5.
Jin J, Ko S, Ryoo C-K. Fault tolerant control for satellites with four reaction wheels. 2008;16(10):1250-1258.
6.
Wisniewski R, Kulczycki P. Slew maneuver control for spacecraft equipped with star camera and reaction wheels. 2005;13(3):349-356.
7.
Spratling BB, Mortari D. A survey on star identification algorithms. 2009;2(1):93-107.
8.
Psiaki ML, Martel F, Pal PK. Three-axis attitude determination via Kalman filtering of magnetometer data. 1990;13(3):506-514.
9.
Chang Y-K, Yun M-Y, Lee B-H. A new modeling and validation of two-axis miniature fine sun sensor. 2007;134(2):357-365.
10.
Greenheck D, Bishop R, Jonardi E, Christian J. Design and Testing of a Low-Cost MEMS IMU Cluster for SmallSat Applications. Presented at the: 28th Annual AIAA/USU Conference on Small Satellites; 2014.
11.
Montenbruck O, Swatschina P, Markgraf M, Santandrea S, Naudet J, Tilmans E. Precision spacecraft navigation using a low-cost GPS receiver. 2012;16(4):519-529.
12.
Office of the Federal Register. FOREIGN AVAILABILITY DETERMINATION PROCEDURES AND CRITERIA. 2015;Title 15 Part 768.7.
13.
Aguirre FH. X-Band electronics for the INSPIRE Cubesat deep space radio. Presented at the: Aerospace Conference, 2015 IEEE; 2015.
14.
Baird D. NASA Tests Atomic Clock for Deep Space Navigation. JPL News. https://www.jpl.nasa.gov/news/news.php?feature=7053. Published 2018. Accessed July 2018.
15.
Cornwall D. NASA’s Deep Space Atomic Clock and Optical Communications Program for PNT Applications. GPS.gov. https://www.gps.gov/governance/advisory/meetings/2016-12/cornwell.pdf. Published 2016. Accessed June 19, 2018.
16.
LP102 Low Power OCXO Datasheet from Bliley Technologies, Inc. (2017).
17.
Kolar JW, Zwyssig C, Kaufmann M, Tüysüz A. High-speed magnetically levitated reaction wheels for small satellites,. In: Anacapri Capri Island, Italy; 2016.
18.
Lee DY, Kuevor P, Culter J. Multi-algorithmic Hybrid Attitude Determination and Control System of the CubeSat “CADRE.” Smallsat 2016. https://digitalcommons.usu.edu/smallsat/2016/S2CDH/2/. Published 2016. Accessed July 14, 2018.
19.
Mishra S. Laboratory Validation of Vision Based Grasping, Guidance and Control with Two Nanosatellite Models. Smallsat 2016. https://digitalcommons.usu.edu/smallsat/2016/S5GuidCont/6/. Published 2016. Accessed July 14, 2018.
20.
Gunter BC, Davis B, Lightsey G, Braun R. The Ranging and Nanosatellite Guidance Experiment (RANGE). Smallsat 2016. https://digitalcommons.usu.edu/smallsat/2016/S5GuidCont/3/. Published 2016. Accessed July 15, 2018.
21.
“Cost-Effective and Readily Manufactured Attitude Determination and Control System for NanoSatellites” by Choueiri, M N; Bell, M; Peck, M A, AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, AIAA SciTech Forum 2018.