Close Support: Staying On Track Satellite Communications On The Move
Mobile Communications Continue to Surge in Growth
Mobile communications are continuing to surge in growth, driven by two key factors the need for greater throughput to support ever more demanding multimedia user expectations, and the need to communicate anywhere, anytime. Fiber, for all its benefits, can satisfy only the first requirement. Meeting the second requirement calls for a wireless solution, and satellites, long denigrated by the fiber community, are the only realistic choice for untethered, long-haul, high-capacity communications from anywhere in the world. In particular, Ka-band satellites, operating at frequencies around 30 GHz, offer much greater channel bandwidths and data throughput than existing satellite solutions at lower frequencies, such as X- and Ku-band.
However, true mobility demands that communications can continue even while users are themselves travelling, for instance in a plane or on a four-wheel drive vehicle. This presents a major problem for satellite communications, because at the microwave frequencies typically used, the signal is directional and the terminal antenna must be oriented directly towards the satellite, for optimal signal transmission and reception. Systems capable of finding the satellite and maintaining the terminal oriented towards it, even from a moving vehicle, are known as On-The-Move (OTM) systems.
Although there are many X-band and Ku-band satellite tracking antennas for OTM applications, only a few have been developed for Ka-band. Most of these are adaptations of lower-frequency systems, rather than designs specifically for Ka-band operation.
The Differences At Ka-Band
What are the differences with OTM systems operating at Ka-band?
First, the Ka-band signal has higher path loss and is more susceptible to atmospheric effects than lower frequency signals. Increased free space loss must be compensated for by higher antenna gain either a larger antenna, or an antenna that is more accurately focused on the satellite.
Second, the wavelength at Ka-band is a mere 10mm in air, and even shorter in typical radome materials such as plastic or fiber. This means the signal can be severely refracted as it passes through the radome, and traditional methods of correctly pointing the antenna must account for the change in angle the signal will suffer as it passes through the radome.
This can be particularly important for OTM applications, as the moving platform will cause the signal to pass through different apertures on the radome as the platform moves. The change in angle of incidence as the signal passes through different sections of the radome can cause refraction in the same way as occurs with an object immersed underwater, shifting the apparent position of the satellite. This is not only a function of motion of the platform, but also of the frequency of the signal, requiring that the tracking signal and communications signal are at the same frequency and derived from the same apparent source.
This effect becomes worse in the high Ka-band frequency band, and is exacerbated by the use of circular polarization in the signal since the effect is also polarization dependent.
With the deployment of the Wideband Global SATCOM (WGS) system by the U.S. military among other Ka-Band satellite deployments, coupled with the demand for continuous Command and Control (C2) on the Battlefield, the demand for OTM terminals operating at Ka-band is expected to increase. With the increase in the number of small terminals deployed, it becomes imperative that these systems use the channel bandwidth available to them efficiently.
EM Solutions, a Brisbane-based advanced manufacturer of innovative microwave communication systems, has developed a dedicated Ka-band On-The-Move Satellite Communications System under Australias Defence Capability & Technology Demonstrator (CTD) Program. The CTD program is designed to investigate and demonstrate technology, and EM Solutions has used the opportunity to implement a number of innovations to improve the performance thresholds for a Ka-band OTM antenna system.
The purpose of this article is to describe some of the challenges in working at Ka-band frequencies.
Mechanical Design To Physical Control System Modelling
At Ka-band, with millimeter wavelengths, mechanical and electromagnetic considerations are even more closely coupled than at X- or Ku-bands. Modelling these effects is critical in the design of the RF system. Achieving high output power levels with solid state components requires careful combining of active power devices no mean feat at such high frequencies! EM Solutions solid state power amplifiers use a novel three-dimensional serial power combining technique that achieves the highest power density (in the smallest volume) in the industry.
However, mechanical design of the stabilization platform is itself a critical task. Design simulation tools will import mechanical designs, created using a 3-D CAD package, into a physical model of the control-system. This modelling extracts mass and inertia properties, joint locations and the physical appearance of the system. EM Solutions used SolidWorks and CosmosWorks for mechanical design, and Simulink and Matlab (MathWorks products) to create mechanical and system control models. This then provided a design platform to optimise and update the tracking system.
The entire antenna control system was incorporated into the model, including non-linear effects such as noise and bearing friction, to yield an accurate physical model. The design process was iterative, with the model being updated to match measurements made on constructed jigs.
Recorded motion data (real or simulated) can be used as an input to the system model in order to simulate the expected tracking performance. EM Solutions used recorded vehicle motion data from a test vehicle (Bushmaster) in the system simulation model. Typical simulation inputs included limits up to 65 degrees per sec for rotational velocity; 300 degree per sec2 for acceleration; and up to 10Hz frequency vibrations. Simulations then allowed the tracking control loop design to be optimised. The following image shows plots of the pointing error magnitude, and motor power consumption of a Ka-band tracking antenna simulation model. The power consumption of the tracking platform can be a critical design constraint in moving vehicles, since their power budget is often limited by other operational systems. In our case, the total power budget of the entire satellite terminal system was limited to 500W. The power consumption of the system was minimized by balancing the RF components at the base of the antenna, to keep its center of gravity balanced in a neutral position, avoiding the need for constant use of the tracking motors to simply maintain the static balance of the system.
Antenna Tracking Mount
Motors and their associated tracking amplifiers need to be optimised as part of the control loop design. This process involves sending test waveforms through each motor in order to characterise the motor response to various inputs. Modifications to the amplifier circuit may be required to meet design control loop parameters of the tracking mount.
Encoders are also required as part of the motor drive circuit and to determine axis position. There are two options for rotary encoders absolute and incremental. Absolute encoders have the advantage that they provide axis position at any instant without needing to move the axis. The drawback is that absolute encoders are more complicated and they are mostly optical, which may not be robust enough for certain applications. Incremental encoders need to have some sort of homing (e.g. a limit switch or another encoder head) in order to determine the absolute axis position. Furthermore, because the control system will generally integrate the pulses coming from the encoder to determine the absolute axis position, any noise that causes pulses to be missed or added will gradually deteriorate the measured position and cause inaccuracies. There are also more options available for incremental encoders compared to absolute encoders, including both magnetic and inductive forms.
Physical Effects Friction and Balance
Friction causes the tracking mount to lose its pointing angle during vehicle motion, so the motors must apply torque to overcome the friction. High friction within the motor and bearings result in the motors having to use more power to overcome the friction. Having some friction in the bearings/motors, however, does alleviate power consumption as the motors will otherwise need to compensate for any out of balance effects. Friction will automatically tend to hold the mount in its present position.
Balance is a critical factor in tracking mount design, as having a balanced system (i.e. one where the axes pass through the centers of mass) can aid in reducing power consumption and increase system performance. In an unbalanced mount, linear acceleration of the vehicle will translate into rotational motion about the axes, forcing the motors to consume power just to maintain the original pointing angle. A more balanced design can avoid this effect.
A further challenge with satellite tracking mounts is the keyhole effect, which occurs when the mount is required to track a satellite at elevation angles approaching 90 degrees from its base (i.e. looking straight up). In some systems, to reduce the overall vertical height, the drive does not permit the antenna to point directly overhead. In this region, such a tracking system would need to rely on its gyros and an external navigation system (referred to as non-closed-loop tracking) to achieve a position as close to vertical as possible, until the antenna is once again able to track the satellite. This blind region may also result in a large movement around the azimuth axis during the reacquisition process.
A system with a two-axis tracking mount, known as an Azimuth and Elevation (Az-El) mount, will experience the worst keyhole effect. When such a mount is facing straight up i.e. at high elevation angles, the azimuth may need to track to a direction which could require a rotation of up to 180 degrees.
The simplest way to prevent the keyhole effect in such a two-axis system is to increase the vertical profile of the antenna mount to allow it to face straight up, in spite of the issue of massive rotation. However in some applications this extra vertical height may be unacceptable, since it increases the height of the total system. To avoid this, more complicated approaches will need to be considered, such as using an additional cross-elevation axis, or using an offset feed. An offset feed, though, simply shifts the pointing problem to lower angles of elevation.
The best solution then is to add a third axis to the system oriented at 90 degrees to the primary elevation axis. This cross-elevation axis may only need limited angular movement to reduce the load on the azimuth axis when the mount is pointing at high elevation angles. The additional degree of freedom in elevation will allow direct overhead pointing with the base able to rotate to an optimal position. Tracking control loops will need sufficient bandwidth on the azimuth axis so it is able to move to a new optimal position before the angle limit on the elevation axes are reached. Adding the second elevation axis will increase the cost and height of the terminal as more mechanical and control system design is required, however it will ultimately result in power savings and less wear on the azimuth axis.
The Value Of Closed-LoopTracking At Ka-Band
Ka-band SOTM operation on the WGS satellite constellation imposes quite stringent constraints on pointing-error control. These constraints are due to a combination of regulatory and link-budget considerations required to efficiently use the available bandwidth on these satellites. For example, during transmission, it is important the beam is pointed directly at the desired satellite, rather than off-centre, where it may leak to another satellite or reduce the desired signal level below its detectable threshold. While the actual pointing-error requirement for a SOTM terminal will depend on a number of parameters, it is likely to be of the order of a few tenths of a degree.
This degree of pointing accuracy is very difficult to achieve with an open-loop tracking system that relies solely on inertial measurement systems to steer the antenna. Furthermore, inertial measurement systems rely on GPS measurements made at a frequency that cannot account for beam refraction through the radome, and are therefore susceptible to radome variations and large offset errors that depend on the angle of incidence and the RF frequency of the measurement signal. Therefore, EM Solutions has adopted a closed-loop tracking system that directly measures the pointing-error using the satellite signal itself.
The pointing-error is the deviation between the desired antenna orientation and its current actual position. It is the control signal that drives the tracking system. A closed-loop control system will attempt to reduce the pointing error to zero. There are many well-known methods for estimating pointing-error. These include:
Phased array (scanning and multi-beam)
All of these approaches would normally rely on the use of a beacon on the satellite.
A. Mechanical Scan
A conventional reflector antenna can be mechanically scanned to estimate the pointing-error. Examples of this approach are conical scan and step-track. Mechanical scanning has two main disadvantages: It requires introducing a deliberate pointing-error, which can reduce the link budget; and it requires rapid mechanical motion so that pointing-error can be tracked during motion of the vehicle.
Monopulse systems are able to estimate the pointing-error without any mechanical scanning and without needing to deliberately mis-point. Monopulse antennas generally have two feeds: One feed has a normal antenna pattern, while the other has a pattern with a sharp notch along the bore-sight. By comparing the signals from the two feeds, the magnitude and direction of the pointing-error can be determined.
While the monopulse is an attractive solution to the problem of determining the pointing-error, it still has a number of disadvantages. Monopulse feeds are generally mechanically complex and so tend to be physically large, making it difficult to integrate into a SOTM terminal. They also require at least two phase matched downconversion chains.
C. Phased Arrays
Phased array antennas have many features that would be beneficial for SOTM. For example, the beam could be steered rapidly, by electronically phase shifting the input signals (the so called inertia-less beam), enabling the use of a high-speed scan without mechanical motion to estimate the pointing-error. Alternatively, a multi-beam phased array could also be configured to operate in a monopulse mode.
Unfortunately, phased array operation at Ka-band presents many technical difficulties. In particular, it is very difficult to share the physical aperture between transmit and receive signals because of the frequency separation between the uplink and downlink bands. This means the phased array antenna must be nearly twice the size of a conventional reflector, if it is to achieve the same system gain. Other challenges are also introduced by use of a phased array, for example ensuring that transmit and receive beams point in the same direction, and proving that regulatory requirements, such as antenna sidelobes, are satisfied for all possible pointing angles. The technical challenges of phased array operation at K-band make the conventional reflector antenna a more attractive solution.
Estimating Noise + Pointing-Error
The pointing-error signal, no matter how derived, is required to steer the beam into the correct orientation from its current position. Thermal noise from the antenna and LNA will cause noise on the estimated pointing-error, no matter which of the pointing-error estimations techniques are adopted.
Noise on the pointing-error estimate will induce errors in the control signal. The antenna control loops treats the noise as a real pointing-error and will try to track it out, thus inducing a real pointing-error. The antenna control loops can only track the noise at frequencies up to their loop bandwidths. Therefore, the antenna control loop acts effectively as a low-pass-filter on the noise.
Narrow-band filtering of the beacon signal is normally required to increase the carrier-noise ratio well above the threshold required for estimation of the pointing-error. Choosing the filters bandwidth is a compromise. It must be narrow enough to reduce noise to a tolerable level, yet it must not be so narrow that it upsets the stability of the antenna control loop and excessively slows down the tracking response. The actual bandwidth required is a function of many system parameters, but values are likely to range from hundreds of Hertz to a few kilo-Hertz.
Actual pointing error induced by noise on the output of the beacon signal processing is only one part of the total pointing error budget. Control systems generally will also still use gyroscopes to correct for higher frequency motion that cannot be corrected quickly enough by the control loop. The following diagram illustrates how beacon noise and gyro inaccuracies combine to result in a total pointing error.
Doppler Shift + Frequency Offsets
The challenges do not end there. Because the platform can be travelling, the satellite beacon, used to generate the pointing error, suffers Doppler shift. Uncertainty in the beacon frequency is consequently quite large. This is due to drift in the satellites own local oscillator as well as the Doppler shifts caused by vehicle motion. The frequency offset can be several hundred kiloHertz, and the Doppler shift can change at a few kiloHertz per second as the vehicle maneuvers. These frequency offsets normally exceed the filter bandwidth typically required in pointing-error estimation. This means that some form of tracking filter is required.
A conventional Phase-Lock-Loop (PLL) tracking filter could be used to follow the wandering beacon signal. However, a Fast Fourier Transform (FFT) based approach is a possible alternative if the pointing-error calculation algorithm is relatively tolerant to small frequency offsets. With either approach, a balance must be reached between the speed, and the accuracy of the filters frequency tracking.
The information above describes some of the options that were considered during the design phase of EM Solutions Ka-band Satcom OTM system. After taking into consideration the performance requirements for operation on the WGS constellation, together with the design options outlined above, it was evident to the engineering team that the most cost-effective and technically low risk option would be to design a tracking system based on a parabolic antenna utilizing closed-loop beacon signal tracking specifically for Ka-Band.
The program has culminated in the production and successful demonstration of a prototype Ka-Band SATCOM On-the-move terminal which provides synchronous satellite link data rates of 2Mbps up and 8Mbps down. EM Solutions has since conducted further development of the terminal to improve the systems RF performance and evolve the terminal from a prototype to a production Satcom On-the-move system. The system will be commercially available in 2012.
About EM Solutions
EM Solutions is a technology provider to commercial and military customers in the telecommunications sector. EM Solutions is a market leader in the supply of Ka-band products to defence and enterprise customers. Their products include LNBs, BUCs and SSPAs, and Fixed Point-to-Multipoint radios based on the WiMAX IEEE 802.16d standard.
The support and cooperation of the Australian Department of Defence through its Capability Technology Development program, which made this work possible, is gratefully acknowledged.