EMWIN Implementation Report

Prepared by

Computer Sciences Corporation

September 28, 2001

I. Introduction and Overview

The Emergency Management Weather Information Network (EMWIN) was established approximately 10 years ago to provide near real-time emergency weather information via relatively low-cost (e.g., approximately $1000/station) L-band receiver systems. The current EMWIN system shares the existing GOES WEFAX L-band transponder to continuously broadcast emergency weather information through the two operational GOES satellites (i.e., GOES East and GOES West) positioned over the United States. EMWIN data is also broadcast from a spare older GOES satellite positioned over the Pacific ocean (i.e., PeaceSat). The low cost of the EMWIN receivers and the importance of the EMWIN emergency weather data has served to grow the population of EMWIN users to well over 10,000.

The power flux density of the current EMWIN broadcasts exceeds the limits specified in current ITU and US regulations. To satisfy these existing regulatory constraints and also adhere to the fundamental principal of conserving scarce RF spectrum resources, the next generation GOES satellites (i.e., GOES N-Q) will have a lower Effective Isotropic Radiated Power (EIRP). To maintain a satisfactory EMWIN performance level (e.g., 1x10-6 BER) as well as the existing 9600 Kbps data rate it may be necessary to utilize improved digital modulation (e.g., Binary Phased Shift Keying or BPSK) coupled with some form of Forward Error Correction (FEC), like Viterbi and Reed-Solomon coding. This potential change in modulation and addition of FEC will require the modification or replacement of existing EMWIN receivers (EMWIN I) with the next generation GOES EMWIN receivers (EMWIN N) . Recognizing the proliferation, popularity and importance of the EMWIN service it is very important to minimize the cost of these new digital receivers. It is also very important to minimize the problems and difficulties for EMWIN users to transition to the new EMWIN service.

This study and report establishes and assesses the plans for implementing and transitioning to the new digital EMWIN service. It begins with an assessment of the basic communications link between the GOES-N satellite and the new digital EMWIN receiver, based on established technical parameters. The study will then establish and assess alternatives for developing an EMWIN digital receiver for a relatively low cost (i.e., less then $2000). Finally, the study establishes and assesses alternatives for transitioning between the current and new EMWIN services.

II. Summary of EMWIN I Requirements, Constraints and Projected Link Performance

The current EMWIN system operates at 9.6 Kbps using FSK modulation. The system shares the existing WEFAX transponder utilizing approximately ½ of the available transponder 12 watt power. The new EMWIN system will initially operate at the same data rate. In future years EMWIN may possibly operate at a somewhat higher (e.g., 19.2 Kbps) data rate as enabled by advances in technology. The EMWIN system will be designed to operate at a Bit Error Rate (BER) of 1X10-6. The G/T will also remain at -0.3 dB to minimize the cost and size of the antenna.

The EMWIN user ground station is currently being designed to operate with the Boeing Satellite Systems (BSS) "projected" minimum EIRP of 43 dBmi, which is certainly feasible since there is always considerable flexibility in designing most ground receivers and adapt them to the receive power levels. The relatively high projected cost of $2000 for this station is based on the BSS projected minimum EIRP level of 43 dBmi. The current design also assumes utilization of the same FEC as LRIT (i.e., R-S Convolutional) although more advanced FEC codes (i.e., Turbo Product and Turbo Convolutional) are also being considered. A brief summary of the background of the EMWIN EIRP requirement and its current status is presented in Appendix A. Appendix A also briefly discusses the impact on the user ground station required to accommodate this relatively low EIRP.

Based on the above link characteristics an EMWIN link budget analysis was performed. The significant decrease in EMWIN N EIRP is planned to be offset by a change in modulation from FSK to BPSK, and the addition of FEC. These two changes combine to provide approximately a 10 dB improvement, which compensates for the significant reduction in EMWIN EIRP. The link budget analysis presented in Figure 1 shows at least a 3 dB margin in all but the worst elevation angle (0°). The system will be designed for a minimum 3 dB margin at the 5° elevation angle. The footprints of GOES East (75° W) and GOES West (135° W) for four elevation angles (5°, 10°, 20°, and 30°) are presented in Figure 2. Additional margin could be available if improved FEC codes, antennas or low noise amplifiers (LNA) are used. This added or surplus margin would be used at the discretion of the government for a number of purposes, including increasing the EMWIN data rate, improving the EMWIN link reliability or availability during adverse conditions (e.g., interference, low elevation angles, scintillation), and reducing the overall cost of the EMWIN user station (e.g., smaller antenna, higher noise receiver, lower cost demodulator).

Elevation Angle

0

5

10

15

20

30

40

60

90

degrees

Frequency

1693.5

1693.5

1693.5

1693.5

1693.5

1693.5

1693.5

1693.5

1693.5

MHZ

Wavelength

0.2

0.2

0.2

0.2

0.2

0.2

0.2

0.2

0.2

Meters

Data Rate

9.6

9.6

9.6

9.6

9.6

9.6

9.6

9.6

9.6

kbit/s

EIRP

43.2

43.2

43.3

43.4

43.5

43.6

43.8

44.1

44.5

dBm

RCVR Pointing Loss

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

0.0

dB

Range

43,450

41,108

40,665

40,140

39,633

38,690

37,859

36,598

35,786

Km

Atmospheric Absorption+Rain

-0.423

-0.400

-0.210

-0.200

-0.100

-0.090

-0.080

-0.060

-0.040

 

FSL

-189.78

-189.30

-189.20

-189.09

-188.98

-188.77

-188.58

-188.29

-188.09

dB

Polarization Loss

-0.20

-0.20

-0.20

-0.20

-0.20

-0.20

-0.20

-0.20

-0.20

dB

Received Power

-147.20

-146.70

-146.27

-146.07

-145.79

-145.43

-145.08

-144.48

-143.83

dBm

G/T

-0.30

-0.30

-0.30

-0.30

-0.30

-0.30

-0.30

-0.30

-0.30

dB/K

Filter Insertion loss

-1.0

-1.0

-1.0

-1.0

-1.0

-1.0

-1.0

-1.0

-1.0

dB

C/No

50.10

50.61

51.03

51.23

51.51

51.88

52.22

52.82

53.47

dB-Hz

Turnaround C/No

49.93

50.44

50.86

51.06

51.34

51.71

52.05

52.65

53.30

dB-Hz

Modulation Loss

-2.30

-2.30

-2.30

-2.30

-2.30

-2.30

-2.30

-2.30

-2.30

dB

Eb/No Theoretical @ 10-6

10.60

10.60

10.60

10.60

10.60

10.60

10.60

10.60

10.60

dB

Coding Gain

6.00

6.00

6.00

6.00

6.00

6.00

6.00

6.00

6.00

dB

Satellite Degradation

-0.71

-0.71

-0.71

-0.71

-0.71

-0.71

-0.71

-0.71

-0.71

dB

Eb/No Available

7.10

7.60

8.03

8.23

8.51

8.87

9.22

9.82

10.47

dB

Eb/No Required @10-6

4.6

4.6

4.6

4.6

4.6

4.6

4.6

4.6

4.6

dB

Margin

2.50

3.00

3.43

3.63

3.91

4.27

4.62

5.22

5.87

dB

 

Figure 1. EMWIN with a 1-meter antenna and no LNA (6-dB coding gain included)

 



Figure 2. GOES East (75°W), GOES West (135°W) Visibility Contours

for Terminal Elevation Angles of 5°, 10°, 20°, 30°

III. EMWIN System Alternatives, Development and Testing

EMWIN N Link Alternatives

The prototype design, development and testing of the EMWIN N receive system should begin in the fall of 2001 and continue throughout most of 2002. The currently planned test windows are November/December 2001 for GOES 12 and June 2002 for GOES 11.

The test results should be a major input for determining the detailed communication characteristics of the EMWIN system including:

1) Modulation (e.g., FSK, BPSK, QPSK)

2) FEC (e.g., RS/Convolutional, KenCast, Turbo codes)

3) Receiver front end (e.g., antenna size, Low Noise Amplifier noise temperature)

The current baseline for these three characteristics are: 1) BPSK, 2) Turbo codes or RS/Convolutional, and 3) 1 meter antenna and 190° Kelvin noise temperature (-0.3 dB G/T). These baselines will be further evaluated and modified as prototype development and testing proceeds. The prototype testing should help determine and validate the expected performance of these communication characteristics. Prototype development has already begun informally and a test model may be ready by November 2001. The FEC technology (i.e., RS/Convolutional) being developed for LRIT is expected to be available to EMWIN at reasonably low cost (e.g., < $100). A relatively low-cost modification ($500) option is also expected to be developed to convert current EMWIN I stations to new EMWIN N stations (e.g., demodulator replacement).

The economic feasibility and cost of the EMWIN N user station will be driven by readily available technology as well as the specific communication characteristics selected by the system designers. To keep cost well below the $2000 objective, low cost technologies are being emphasized. For example, only FEC decoders below $100 are currently being considered.

EMWIN FEC alternatives (e.g., R-S/Convolutional, "KenCast", Turbo codes)

An initial and preliminary evaluation of the three FEC alternatives (RS/Convolutional, KenCast, Turbo codes) introduced above has been completed. All three are projected to achieve the FEC cost objective of <$100 and, in fact, could even be below $50. The RS/Convolutional option is similar to the FEC plans for LRIT that were driven by the need to comply with the CGMS LRIT specification. This option may benefit from the ongoing LRIT development effort, which will ensure the availability of competitive and relatively low cost RS/Convolutional receiver systems. The KenCast option has been coordinated with the commercial vendor of the same name and is still under evaluation. It is advertised as a very effective and versatile FEC, but currently is marginally rated for two reasons: 1) it is a proprietary FEC and therefore has the inherent disadvantage of requiring a licensing fee, 2) KenCast has not yet provided much assistance in helping to evaluate the effectiveness of their approach. Currently the highest rated approach is Turbo codes, for two reasons: 1) it utilizes an advanced Turbo code that is about 2 dB more effective than the more traditional RS/Convolutional codes currently planned for LRIT, and 2) it is probably a relatively low-cost (<$50) easily-implemented software or hardware solution.

EMWIN receiver development alternatives and assessments

The EMWIN receiver design and development effort has begun with the preliminary definition and validation of the next generation (EMWIN N) requirements and constraints. Ongoing OSD engineering and development efforts will continue through close cooperation and coordination among NOAA NWS and NESDIS groups, including OSD and OSO. OSD will continue to consult and coordinate with key industry vendors to ensure the satisfactory performance and availability of low-cost EMWIN N receiver stations. Major alternatives currently under consideration focus on a specific modulation (e.g., FSK, BPSK, QPSK) and FEC (R-S/Convolutional, Turbo code, etc.). Tradeoffs between cost, performance and capacity (i.e., data rate) will be the major criteria in making final decisions. The current baseline is BPSK combined with an available Turbo code product. This baseline will continue to be tested,evaluated and coordinated throughout the remainder of 2001.

 

 

IV. EMWIN Transition Alternatives

The transition from existing EMWIN I broadcast services to the new EMWIN N broadcast must take into consideration at least three factors:

 

Three basic transition alternatives were established and evaluated:

1) Providing a third overlapping satellite (e.g., GOES Central) to transmit EMWIN N during a specified transition period (e.g., 1-2 years).

2) Immediate and total transition to the EMWIN N service for either GOES East or GOES West, followed by a later (e.g., 1-3 years) total transition of the other satellite.

3) Time sharing between EMWIN I and EMWIN N on individual spacecraft for a limited time period (e.g., 6 to 12 months) followed by a total transition.

The preferred recommendation for the transition to EMWIN N services is Alternative 1. This alternative would allow the maximum flexibility for EMWIN users since they would continue to have the availability of both broadcasts (EMWIN I from GOES EAST and GOES WEST and EMWIN N from GOES Central) full-time during the transition period. The major drawback of this alternative is the resources required to operate a third geostationary satellite. The additional resources to be considered and estimated include:

Approximately one full-time equivalent (FTE) OSO staff member is projected to be required for the more "limited" operations of the spare GOES. The antenna and other ground systems at WCDA and SOCC, projected as requirements, are now currently available.

Alternative 2 (i.e., total and sudden transition for individual satellites) is undesirable from the user's perspective. It would necessitate a sudden and difficult transition phase for WEFAX users. Most users, especially DOD users, have a strong need for a relatively long transition period when both the old and new services are provided simultaneously. Users generally require a "reasonable" transition period to plan and provide for the acquisition, implementation and validation of any new earth station like EMWIN. Vendors also need a "reasonable" transition period to plan, develop, market and distribute their upgraded EMWIN systems. A sudden turn-off of EMWIN I and turn-on of EMWIN N would deny both the user and vendor of fair and reasonable transition process and could very well discourage users and vendors from supporting the EMWIN N service. This alternative provides virtually no transition period.

Alternative 3 is the least desirable. This alternative can be described as a period of parallel operations for each of the two GOES satellites, where both EMWIN I and EMWIN N services would be simultaneously broadcast to a timeshared GOES I-M transponder for a specified transition period, followed by a full and permanent transition to full EMWIN N broadcasts. This alternative was evaluated by NWS and is summarized in Appendix B. The current 90% duty cycle would require almost 50% reduction in content. There would also be an increase in the latency of the messages. Synchronization and flow control of the two data streams could also be a problem.

All three transition alternatives are currently being planned for the operation on the existing GOES I-M spacecraft to permit the establishment of firm transition dates that are not dependent on the GOES N launch and operational dates. Ongoing technical assessments and tests are currently being performed to validate the ability of the EMWIN N broadcast to coexist with both the existing WEFAX signals and the new LRIT signals without interference. Recent compatibility testing between the LRIT and EMWIN I signals were encouraging or positive. However, testing between EMWIN N and both WEFAX and LRIT was not complete or conclusive. Therefore, additional testing is still required.

Considering the above factors and recommendations, the three alternatives are being further evaluated based on various program considerations.

V. Conclusions and Recommendations

This study has concluded the development, implementation and transition of the next generation EMWIN system (EMWIN N) is entirely feasible at a "reasonable" cost (e.g., <$2000). Additional study, development, and testing efforts are required to establish the final system design in preparation for both a final system specification and an EMWIN N receiver specification.

The recommended phases or steps to continue the EMWIN design, development and implementation are:

1) Complete preliminary system design alternatives and the definition of specific tests and evaluation procedures to select specific alternatives for a final system design.

2) Perform required tests to determine and validate design characteristics.

3) Complete of a final system design specification, including a specifications for all major system elements (i.e., NWS, SOCC CDAS receiver station).




Appendix A. EMWIN EIRP Requirements Overview


The GOES-N EMWIN requirements are presented in Performance Specification S-415-22 (P Spec) dated August 26, 1997. Two unrelated EMWIN EIRP requirements are specified. The first requirement is specified in paragraph 10.2.1.8 and is based on BER performance and link requirements. The second requirement is specified in paragraph 10.2.6 and is based on providing the "maximum allowable EIRP" as determined by ITU-R PFD constraints. Without any stated reason for this second EIRP requirement (i.e., "maximum allowed") it is assumed that the original intent of the requirement included such standard reasons as: 1) providing added link margin for digital links, which tend to fail completely and catastrophically whenever the signal falls below some minimum level, 2) providing the flexibility to increase the data rate at some time in the future, 3) extending the coverage area or footprint, and 4) increasing the potential and flexibility for reducing the cost and complexity of the user station. After much detailed technical discussion about this second requirement and serious concern about contractor compliance in meeting the mutually agreed upon value of 49.4 dBmi, this second requirement seems to have been contractually eliminated with little regard for the consequences.

The actual/contractual EIRP value that would satisfy the first requirement was negotiated during the last six months of 2000 and as of mid December 2000 the differences seem to have been reduced to a few dB, somewhere between 43.2 dBmi and 45 dBmi. In addition to uncertainty about other link budget factors, there is still some uncertainty over the type and performance of the forward error correction (FEC) code (e.g., convolutional, Reed-Solomon) to be employed by EMWIN.

The GOES N EIRP requirement and associated acceptance test criteria should be determined by the GOES N Performance Specification (P Spec) and should not be impacted or adjusted by any characteristics or changes to the EMWIN ground station. Any benefits or disadvantages associated with the actual design or specifications of the EMWIN ground station are the responsibility of the NOAA, including all associated costs and benefits.

In response to repeatedly expressed concerns about the increased cost of the new EMWIN user station, BSS was tasked in June 2001 to assess the feasibility and cost of meeting the "maximum allowed" EIRP value of 49 dBmi. This six dB increase could allow the cost of the new user to be decreased by eliminating or reducing the need for some of the performance enhancing characteristics such as FEC or low noise amplifiers (LNA).

A preliminary EIRP impact assessment was performed. The higher cost of the "minimum" EIRP, 43 dBmi, EMWIN I user station is estimated at $1500 to $2000 and assumes a BPSK demodulator, a concatenated R-S/Convolutional FEC decoder and a front end receiver with a G/T of -0.3 dB. One of the potential impacts of the 6 dB EIRP increase currently being studied, would be the reduction or elimination of one or more of the following non-trivial features: 1) FEC (i.e., Reed-Solomon and ½ rate Convolutional FEC), 2) LNA, or 3) BPSK modulation. Elimination of some combination of these three features could reduce the estimated $2000 cost of a new EMWIN user station by approximately $300 to $900. It could also make it more feasible or practical to retrofit or modify existing EMWIN user stations for an estimated cost of $500 to $1000. The above cost estimates could be validated by developing and testing specific prototype stations. These prototypes should be tested under a variety of conditions from best to worst with respect to signal strength and interference.

Appendix B. NWS Assessment of Alternative 3

(Time Sharing)


The working group planning transition of EMWIN from GOES I-M to broadcast on the GOES N-Q series has proposed the use of a time sharing technique on existing operational broadcasts to provide a users the opportunity of receiving and processing a simulated GOES N-Q transmission in advance of the operational broadcast.

Staff members of the NWS Telecommunication Operations Center responsible for the operation of EMWIN and staff members of the Performance and Awareness Division responsible for servicing the users of EMWIN have analyzed this proposal. Their analysis has raised the following issues.

The current EMWIN data stream requires more than 90 percent of the available bandwidth on the current broadcast. The data stream is carefully managed to ensure that all products satisfy current requirements. Time sharing would necessarily reduce the content of the data stream; 50-50 sharing would reduce the content by half. We would anticipate a strong user objection to reducing data content of this operational data stream by half - even for a good cause like a graceful transition.

The switch between modulation schemes requires time to re-synchronize the receiver systems. The result is an effective loss of bandwidth each time the modulation changes. The current receiver system requires about 30 seconds to re-synchronize and thus has been designed with a 120 second buffer to accommodate momentary loss of signal, re-acquisition, and re-synchronization. Occasionally, when the EMWIN signal is interrupted at the receiver system due to an eclipse, the receiver system will not re-synchronize until the electrical power is cycled, rebooting the system.

The current EMWIN broadcast operates at 9600 bps yielding an effective throughput of about 750 text characters per second in packets. Most watches and warning are relative short, a few thousand characters, but some exceed 6000 characters. Most NWS-originated messages are issued singly but messages can be issues simultaneously by multiple stations. A worst case scenario would be a string of five 6000 character messages taking about 40 seconds to broadcast. In a time sharing mode with frequent switching this scenario would result in additional delays.

The current EMWIN broadcast system has no provision for flow control originating external to the system that we believe would be required by the time-sharing system manager. Without flow control, packets would be lost during the cyclic time sharing period.

The proposed time-sharing alternative significantly reduces the effective bandwidth, may add significant delay of critical information, and is not technically feasible unless flow control can be retro-fitted to the existing EMWIN broadcast system. These issues make time-sharing the EMWIN broadcast between two modulations a very unattractive transition alternative.