Site: Third Ka-Band Utilization Conference
Date Visited: September 15-18, 1997
WTEC: W. Brandon (report author), K. Bhasin, R. DePaula
This conference incorporated material timely to this (1997 NSF) study on the rapidly emerging use of Ka-band for advanced satellite communications, and hence in reported in summary form, analogous to a site visit. It is important to note that this was the third annual meeting, but the report deals only with the 1997 conference. The meeting presented 84 technical papers on proposed systems, intersystem interference, component and other technology advances, data transmission protocols, propagation, markets, regulatory and policy framework, and other uses of Ka-band (such as tracking and data relay satellites (TDRS), deep space and Skystation, a balloon-borne repeater concept).
The explosion in communications, overcrowding of lower bands, and success of the NASA ACTS, ITALSAT and other international initiatives have provided the impetus for proposals for a large number of national, regional and global systems in Ka-band (30 GHz uplink/ 20 GHz downlink). It is appropriate to summarize the proposed systems as a baseline to this report. Table 3.2 on page 34 provides a partial description of some representative systems in terms of parameters chosen to relate to the topics in this summary report.
A special case exists in the realm of data relay satellites which are employing Ka-band intersatellite links and high data rate downlinks. Unlike previous block versions of TDRS, the present program (TDRS H, I, J) does not incorporate commercial band transponders in the payload. The uplink band (GHz) and the downlink band (GHz) are adjacent to the commercial communications band and components developed for either application may be useful in the other. Europe and Japan also operate data relay satellites. Although this was not the intent, the continuing need for tracking and data relay operations may provide an avenue for component and sub-system development for improvement of Ka-band communications satellites.
Most of the proposed systems are aimed (wholly or in part) at multimedia applications, meaning e-mail, Internet connectivity, video distribution, graphics, bandwidth on demand, data broadcast, intranetworks, financial and other data services for business. Eight papers focused on aspects of this new market and a dozen other papers reviewed the topic in some sense; this section is a synthesis of the material on markets from the entire conference. The dominant single impression regarding the multimedia market is that its size, and temporal and spatial distribution are all speculative in some degree. Yet, the ability to mix graphics, text, motion video and sound in a personal computer enables both the conception of satellite delivery and the conviction that a multimedia market is endemic to the fact of computers. Three market segments are generally identified: corporate or business, video distribution, and home or personal user. The other major application is direct access to high data rate transmission (e.g., 55 Mbps or 155 Mbps) using ATM.
The corporate market is the most likely driver for multimedia in Europe (videoconferencing and broadcast, data broadcast and file transfer; telework, telemedecine, teletraining and disaster relief); of 250,000 terminals estimated for the total market, 65% are corporate. This market will develop along different lines and lag behind that of the United States.
In Europe, the consumer or home user market is perceived to be strongly dependent on terminal price and bundling of interactivity and appealing programming for different European cultures and languages. Because of present success of the Internet (2.5 million households to grow to 23 million subscribers by 2003 in Europe) the satellite multimedia market is predicted to grow around Internet-like access/applications. A high speed return path is not considered a value added feature. Terminal price threshold may be $500 but cost may be $700 to $1,000, requiring an intensive subsidy marketing campaign. That the market may develop in unforeseen ways, has led to flexible design to adapt to market conditions (e.g., array antennas to allow distribution of bandwidth and power into high traffic areas, reposition coverage on orbit, etc.) raising interest in phased array antennas.
Video distribution includes distributing high definition digital motion pictures to theaters and enabling distributed high definition video (HDV) production (i.e., providing a virtual studio environment for the director who assembles scenes transmitted from remote teams).
Although many references were made at this conference to "market studies," only one paper actually outlined a market study. The background is the perception that introduction of satellite multimedia and continuing spread and improvement of terrestrial capability will result in heightened user expectation of access to these services beyond the coverage of fixed systems. Thus, whereas users will ultimately desire mobile and transportable service, the planned systems are aimed at fixed terminals. This construct leads to the notion of a system aimed primarily at mobile users, probably with laptop computer based terminals of low cost. Larger and fixed users would then benefit from reduced size and cost of terminals. A range of terminal types from handheld (<$1,000), high rate transportable ($2,000), to fixed high rate (>$10,000) were defined to a range of users. Some of the details learned about applications are as follows:
Some amplifying applications concepts from other papers are: in business-banking, entertainment, aerospace and utilities; for universities-distance learning, tele-astronomy, telecommunications protocol evaluation; in medical services and telemedicine-client-server relational database/patient records heterogeneous distribution, and telemammography (56 million U.S. distributed patients needing access to ever centralizing medical specialization).
The gross domestic product (GDP) was cited as one of "the most important and stable indicators of economic development" of a country (and used to predict traffic and even revenue). Studies also used population centers as market predictors. Nevertheless, one view was that the rural telephony market was likely to be the larger near term market for Ka-band, rather than multimedia. This is based on the concept that in some regions, the proposed MSS systems (aimed at telephone applications) will not have adequate capacity to service the burgeoning market (e.g., China).
Related to the studies and market surveys are the demonstrations and experiments conducted by several organizations.
Alenia Aerospazio is experimenting with fully satellite-based multimedia for telemedecine, tele-education, tele-work, diffusion and newspaper distribution, and Internet access using a Ku-band uplink (Hot Bird) and a Ka-band downlink (ITALSAT), under the European Commission's ISIS and ESA's IMMIS projects.
An experimental distributed network using satellite and terrestrial networks for tele-education and telemedicine in Italy reported limited initial results from 400 hours of testing. A major finding (in tele-educaton) is that a symmetric configuration of 384 kbps represents a good compromise between service quality and satellite capacity. In the unsymmetric case, the best results required a return link rate of 128 to 256 kbps with 384 kbps in the broadcast channel, thought to identify a basic difference between ordinary video teleconferencing and tele-education. (This is in contrast to the above finding that the high data rate return link was not considered a value added feature to the home user.) With the particular protocols used, GEO roundtrip delay did not impact (user perceived) service quality.
In February 1997, 69 nations signed an agreement on international trade "in basic telecommunications services" that is viewed as liberalizing the international regulatory environment, including satellite service. However, reservations or caveats recorded by various national trade administrations could impact multimedia satellite service provision. Lack of a definitive agreement is attributable to (1) the size of the communications commerce (about $600 billion in 1996) and (2) the fact that government monopolies supply services in many countries. Signatories account for 90 percent of the revenues in world telecommunications services.
The agreement requires signatories to accord all World Trade Organization member states most favorable status, i.e., not to treat foreign suppliers differently from nationals. A problem area exists in direct broadcast satellites. The United States has traditionally viewed this as a form of "basic telecommunications" whereas European Union nations treat it as "broadcasting." The distinction is viewed as important due to cultural concerns about content of audio-visual material, creating the tradition of separately treating this area. The EU filed a statement that "commitments...do not cover the economic activity consisting of content provision which require telecommunications services for transport." This suggests that a real agreement has not been reached on regulatory treatment of multimedia as "basic telecommunications." One paper concluded that the uncertainty about this fundamental definition would have implications for Ka-band systems.
Present filings for Ka-band systems appear to have exhausted nearly all the available spectrum. Data presented suggested that many GEO satellites have requested the same orbital position and frequency assignment. The use of large numbers of spot beams and various polarization schemes will allow some degree of overlap in spectrum and orbit slots. No suggestion was made at the conference regarding additional frequency allocation.
A part of the Ka-band (25.25 - 27.5 GHz) is used for earth exploration downlinks on a secondary basis, but will eventually be overcrowded. Upgrading this to primary status is a WARC-97 agenda topic. Hopefully, success in this area will not be balanced by a loss in the communications satellite bands.
In addition to satellite array antennas, onboard processing is a major challenge in Ka-band systems; it is required in order to realize the use of VSATs and spot beams with frequency reuse.
A new product scheduled to be available (engineering model units) in May 1998 is a programmable or digitally controlled variable analog SAW filter. This device can be used to split a rf bandwidth into separate channels for connection to a switch or demodulator.
Concepts for using the high data rates possible in Ka-band vary from switching high burst rates (from different uplink beams) directly into downlink TDM data streams, to applying bursts to demodulators for processing and reassembly into a TDM stream for downlink retransmission. Concepts of uplink burst rates vary from 32 Mbps to 155 Mbps.
CRL described a conceptual design for a 2003 launch Gigabit Communications Technology Satellite that would have 2 fixed and three scanning spot beams. Uplink rates could be 51.84 Mbps (3 channels/beam) or 155 Mbps (single channel/beam). A microwave switch matrix having a microwave center frequency to allow large bandwidth, would switch these channels between uplink and downlink beams. For baseband demodulate/remodulate processing, two approaches are being planned: a fully digital and a hybrid analog/digital technique. The all digital approach was thought to be currently limited to about 60 Mbps and the hybrid approach to about 200 Mbps. Concatenated codes would be used to achieve an error rate of about 10-10 needed for ATM.
ESA has previously developed an onboard processor incorporating a non-blocking switch capable of switching 8,000 channels @ 32 kbps or 256 kbps. Studies recommended 256 channel switching 32 Mbps streams for satellite applications, which was based on writing inputs to RAM. An ATM switch was being implemented for use on a satellite, based on terrestrial switch technology. An Alcatel 1000AX switch, implemented using 74K gate ASICs of 0.6 µm CMOS technology, is being adapted for satellite use by Bosch Telecomm. A feature of the satellite version of the ATM architecture is the splitting of the termination unit function into two parts, placing only a portion in the satellite. This defines a radio link (ground to space), but appears to imply, as stated directly in several other papers, a preference for not using a "pure" ATM for satellite transmission.
Asynchronous Transfer Mode (ATM) was widely accepted for use in the multimedia satellite systems. ATM can support diverse requirements from a variety of traffic sources and provide flexible, efficient switching and transport. The primary problem with satellite ATM is due to the "latency" or the propagation delay time for a satellite link. ATM was devised for fiber optic transmission (very low error rates and low latency). When TCP/IP is used via satellite ATM, three issues appear: latency, slow start and congestion avoidance algorithms and packet loss recovery.
The "latency" problem is really a result of the fact that TCP/IP requires periodic acknowledgement from the receiver, requiring two round trip times in the satellite case. The standard TCP/IP limits transmission to 64 kbytes before receipt of an acknowledgement, or:
(Data Rate) x RTT = 64 kbytes = 512 kbits
where RTT is the round trip time. It can be appreciated that extending the allowable amount of data before acknowledgement would alleviate this difficulty. Such an "extension" of the ATM protocol is defined and must be employed for satellite communications. It is also true that very high data rates (e.g., in fiber) suffer from this problem even with shorter delays. Therefore the issue is shared by both fiber and satellite communications.
The ATM Unspecified Bit Rate (UBR) service category is expected to be used by a wide variety of applications and traffic sources, and hence is of importance for satellite transport. Similarly, TCP/IP used on the Internet is of great importance to satellite transmission. A simulation was reported of performance of TCP over ATM/UBR using three values of latency or Round Trip Time (RTT) to simulate ground connection, multiple LEO hops, and a GEO satellite round trip (5, 100, 275 ms). The simulation measured the throughput efficiency versus buffer size, for various numbers of traffic sources. With TCP selective acknowledgment, about 0.5 RTT to 1.0 RTT is sufficient to provide 98 percent throughput to infinite TCP traffic and a large number of sources.
One system design study, based on interoperability with UMTS, suggested use of an ATM interface (effectively a termination) at the user satellite terminal, so that the satellite access could be optimized for better performance.
ATM transmission experiments (156 Mbps, 200 MHz bandwidth) using N-Star were reported by CRL together with plans to test IP with ATM and satellite-fiber connection.
Improved components and devices were reported in two sessions (13 papers) but notably within a number of papers on other topics. This data is summarized and ordered here by components for satellites (antennas, SSPAs, LNAs, ISLs, processing repeaters) and ground stations/terminals.
Results of a trade study comparing a cluster of horns feed versus a 256 element array feed, in the context of a gigabit/sec downlink with a few fixed beams and 3 scanning spot beams, was presented (Toshiba). The horn feed was driven by a high power TWT with power switched to the appropriate horn for a given beam; the array used one SSPA per element with spatial power combining. Comparison was made between phase-only control and combined amplitude and phase control, with results suggesting that for scan angles greater than 5 degrees, gain of 50 dB can be obtained with phase-only control (uniform amplitude) for scan angles of less than 10 degrees. A dual offset-reflector with a 4.05 meter main reflector configuration was recommended. A related study (reported by CRL and Mitsubishi) recommended a 2,200 direct radiating element (selectively populated or "thinned") array, to be realized with MMIC devices using pseudomorphic HEMPT technology, for SSPAs and LNAs to achieve the 72 dBw EIRP and overall noise figure (NF) of 4 dB. ESA is evaluating a defocused array feeding a Gregorian reflector antenna. Ka-band phased array antennas were noted as "in development" for Astrolink and TDRSS; ARTEMIS and DRTS have developed precision high gain reflector antennas for tracking LEO spacecraft from GEO. An active array is under investigation by a Canadian consortium; a scalable 7 element subarray at 30 GHz has been constructed. Elements have stripline feeds with integral filters driving a dielectric resonator within a circular cavity capable of 5 percent bandwidth (circular polarization). Most systems already in development employ fixed horn (cluster) feeds and large reflectors.
Two innovative antennas were reported. One was a "reflectarray" consisting of a flat surface array of patch radiators illuminated by a horn feed, thought suitable for attaching to a microsatellite. At the other extreme, a 130 beam phased array composed of 736 elements, each with seven horns illuminating one lens, and the elements attached to a spherical surface. The latter antenna was being developed for a balloon-borne repeater (SKYSTATION).
Various LNAs (for satellites) were reported, with the Mitsubishi paper reporting a near ultimate performance at 1.0 dB (world lead). Reported performance specifications include the following:
ESA has developed a Peltier cooled front end that could employ unscreened (i.e., not selected for low NF) p-HEMTs with NF up to 3 dB, but yielding a 1.5 dB NF amplifier.
A microwave power module-consisting of a linearizer, wideband MMIC driver, a short TWT (miniaturized) and an integrated power supply-is being designed as a potential replacement for the entire transmit rf chain. It is expected that by mid-1998, an MPM will be available at 108.8 kg, 30 to 130 watts, with power consumption reduced by about 400 watts (to 5,383, compared with 5,789 watts current at the time of this conference). The MPM is being designed to accept TWTs from AEG, NEC, and TTE.
Some progress in SSPAs for both satellites and earth stations was reported by Mitsubishi. For the 20 GHz satellite SSPA, a Source Island Viahole FET was developed for low thermal resistance (18°C/W) and low parasitic inductance (1.6 pH). Using this device, a MMIC was developed; combining 16 MMIC chips produced 43.8 dBm at 21 GHz, a world record for an SSPA at this frequency.
For 30 GHz, a high power MMIC approach with superior transistor development was employed to reach a 28.5 dBm output at 25 percent power added efficiency, intended as a driver stage. The transistor was a pseudomorphic HEMT double hetero-layer structure with T-shaped gate, produced by electron beam lithography. Power levels were indirectly reported in association with VSATs in a number of papers. The most often quoted value is 1 watt. A 1 watt SSPA was mentioned in the ISIS (EC) project; and 2 watts for IMMIS (ESA). Two watts was also developed for a VSAT power supply combining 4 MMIC chips, each using two devices.
Several novel antennas for mobile platforms have been developed by CRL, including a waveguide slotted array and torus reflector, both mechanically steered in azimuth, and an active phased array. The slotted array requires separate antennas for transmit and receive. The transmit antenna has 16 leaky waveguides with 192 cross slots (for CP); the receive array has 12 waveguides and 142 cross slots. Both are 250 mm diameter, and weigh about 4 kg. Transmit gain is 25.2 dB at 30.8 GHz with 5.7° azimuth beamwidth and 17° elevation beamwidth. Receive gain is 22.5 dB at 21.03 GHz (6.5° x 19.5º°). The design takes account of the fact that the elevation angle to the COMETS satellite does not vary appreciably within Japan, and the antenna need only be steered in azimuth.
The 21.03 GHz active phased array is fully steerable. Due to budget limitations, only a receive array has been developed. It has 168 rectangular microstrip radiating elements, with integral 4 bit MMIC phase shifters. The G/T is - 6.8 dB/° K.
The torus is a dual reflector with parabolic torus main reflector (200 mm diameter) and ellipsoidal sub-reflector. The beam is steered in azimuth by rotating the sub-reflector, and no rotary joints are required. Intended for 44 GHz receive and 47 GHz transmit experiments, the antenna also assumes a small variation in elevation angle, with elevation beamwidth of 12° and elevation beamwidth of about 10°. The complete antenna is 425 mm in diameter and 400 mm high, and weighs 16 kg.
Motivated to achieve a low cost modem for two way multimedia applications, CRC reported development evaluations of alternative modems for the suitcase terminal developed for the ACTS program. Direct I&Q mod and demod were investigated for Ka-band and compared with L-band (implying separate up or down conversion). Direct L-band mod/demod performance was within 2 dB of theoretical and similar to 70 MHz IF modems, at 1.544 Mbps. At 20 GHz, the direct I&Q modem was again within 2 dB, but at 30 GHz, the discrepancy was larger (about 4 dB).
CRL is developing a portable Ka-band terminal with a 64 element patch array for both transmit and receive (~17 dB antenna gain, 1 watt, 290K system temperature, intended to support 2.4 kbps).
Fifteen papers (plus three in a special evening workshop on ITALSAT results) addressed propagation measurements, rain models, automatic fade compensation and methods of accounting for rain in system design. A considerable amount of new rain fade data has been acquired by researchers throughout the world. A comparison of ten different rain models incorporating 21 station-years of new data with 86 station-years of prior data at Ka-band produced some surprises. The DAH rain model (named for authors Dissanayake, Allnut, Haidara) produced better results (RMS error relative to measured values average over the set of stations), for example, than the Crane model for U.S. locations. (The Crane model is said to be under revision.)
Gaseous vapor absorption can be critical at Ka-band, especially for small antennas with low margin links. Some discussion ensued about whether it was useful to separate vapor, rain, and wetting of the antenna and feed losses. The last is really not a propagation loss, but is nevertheless real. (Wet antenna reflector, feed or radome losses up to 6 dB have been observed and may corrupt propagation loss data.) A method of combining independently estimated losses was described. Automatic compensation algorithms were described. The availability for various services is determined by the margin and local fading.
Availability of 98 to 99.5 percent is typical of the performance promised by the proposed systems. No experimental work has apparently been performed to relate availability to customer satisfaction.
Ka-band (32 - 34 GHz) is used for deep space, interplanetary missions. Transponders, which receive and retransmit a high power ground transmission, are used to track deep space trajectories. Since the transponder uses LNA, up and down converter, and SSPA or TWT, and efficiency, light weight, and low power consumption are important, transponder developments for deep space tracking may have important technology spillover benefit to communications satellites. Major weight reduction projects are underway.
Proceedings of the Third Ka-band Utilization Conference. 1997. Genoa, Italy: Istituto Internazionale delle Comunicazioni.