[Top] | [Previous] | [Next]

B-2. Telemetry Assumptions

The reference architecture for satellite TT&C operations assumes a fairly comprehensive telemetry processing capability that can support the performance range required for all SCN satellites. The primary performance factors are data rate, data expansion, data archiving scheme, and processing load. To limit the extent of the performance analysis accomplished within the timeframe required of this study, assumptions had to be made regarding each performance factor.

Data Rates. SCN telemetry rates range from 75 bps to 5 Mbps and the number of telemetry streams per satellite can be one or two. Within the current SCN satellite inventory, the most stressing satellite configuration includes a 128 Kbps SOH stream and a 1.024 Mbps mission/sensor stream, which includes up to 128 Kbps of SOH data (total of 256 Kbps SOH data). For the near future, higher mission/sensor stream rates can be expected, possibly up to 150 Mbps, but SOH data rates are expected to remain the same or drop, due to greater onboard data processing.

There is, however, one factor which mitigates the rate-driven Telemetry processing load. In the TT&C architecture environment, which primarily supports satellite SOH data evaluation, the processing of mission/sensor data, for example to derive an infrared signature, can be excluded from the TT&C system load and relegated to a separate subsystem. Thus, to bound the Telemetry load problem, a maximum SOH rate of 256 Kbps and a nominal SOH rate of 8 Kbps will be assumed.

Data Expansion. The data expansion performance factor has at least two possible cases. An example of a worst-case condition, based on a stressing data capture and monitoring environment, results in greater than a 21:1 data expansion, based on the following:

* Telemetry words average eight bits, with each telemetry word representing one measurand.

* Each 8-bit telemetry word is assigned a 16-bit tag at decommutation, resulting in a total of 24 bits/measurand (3:1 expansion).

* Time is processed as a 32-bit word plus a 16-bit tag, and an 8-bit status value plus a 16-bit status tag, for a total of 72 bits; time occurs as frequently as required based on user requirements and data collection implementation; the frequency of time tagging can range from once per masterframe to once per measurand; thus time can add anywhere from 72 bits per masterframe to 72 bits per measurand to the stream expansion.

* EU conversions transform 8-bit telemetry words to 32-bit words (i.e., real numbers), with a new 16-bit tag, resulting in a total of 120 bits/measurand (15:1 expansion).

* Quality indicators are associated with each measurand, but the number of additional bits required per measurand is implementation-dependent; a worst case might be 8 bits of quality data per measurand plus 16 bits tag, for a total of 144 bits/measurand (18:1 expansion).

* Alarms are directly associated with each measurand and are 8 bit words with a 16-bit tag, resulting in 24 additional bits for each measurand's information packet and a total of 168 bits/measurand (21:1 expansion)

A more probable case for the SCN satellite control environment reflects a nominal telemetry wavetrain structure and results in a 9:1 data expansion, based on the following:

* Telemetry words average eight bits, with each telemetry word representing one measurand.

* Each 8-bit telemetry word is assigned a 16-bit tag at decommutation, resulting in a total of 24 bits/measurand (3:1 expansion).

* Time is processed as a 32-bit word, plus a 16-bit tag (no status or status tag) and occurs once per mainframe, adding 48 bits per mainframe (3:1 expansion + time).

* EU conversions transform 8-bit telemetry words to 32-bit words (i.e., real numbers) with a new tag, and the raw word is discarded, resulting in a total of 48 bits/measurand (6:1 expansion + time).

* Quality indicators are associated with each mainframe (not measurand), adding 8 bits of quality data plus 16 bits tag per mainframe, for a total of 24 bits/mainframe (6:1 expansion + time + quality).

* Alarms are directly associated with each measurand and are 8 bits with a 16-bit tag, resulting in 24 additional bits for each measurand's information packet and a total of 72 bits/measurand (9:1 expansion + time + quality)

If the above expansion factors (the more probable case) are varied to reflect a non-nominal telemetry wavetrain structure (i.e., a large number of discrete measurands), the average measurand length could be reduced to 4 bits. The resulting expansion factors would be:

* Decommutation: 4-bit words with 16-bit tag, 20 bits/measurand >> 5:1 expansion

* EU conversion: 4-bit words to 32-bit words with 16-bit tag, 48 bits/meas >> 12:1 expansion

* Alarms: 72 bits/measurand >> 18:1 expansion

For the Telemetry loading analysis, a nominal wavetrain structure with both nominal and non-nominal expansions will be considered. For the nominal case, a data expansion of 3:1 for the mission data stream and 9:1 for each of the two possible SOH data streams will be assumed. Time and quality data will not be included due to their relatively low composite throughput of 72 bits/MF. This relationship can be described by the following equation:

Nominal Throughputmax = 3M + 9(S1 + S2)

where: M = mission/sensor telemetry rate

S1 = SOH telemetry rate

S2 = embedded SOH telemetry rate, e.g. in mission stream

For the non-nominal expansion case, a data expansion of 3:1 for the mission data stream will remain since mission data does not contain discrete measurands and only tagged, raw mission data words are required for further processing. However, for SOH data, the 18:1 expansion will be assumed, with an output of tagged, EU-converted, and alarmed measurands. This relationship can be described by the following equation:

Non-Nominal Throughputmax = 3M + 18(S1 + S2)

Data Archiving. The data archiving performance factor varies according to the archiving scheme selected for implementation and the location of the storage device. If raw, time-tagged mainframes are archived, then the additional throughput requirement is the raw telemetry rate plus the 48-bit time tag per mainframe plus a small overhead for packet formatting. On the other hand, if raw, decommutated measurands are archived, then the additional throughput requirement is increased due to the 3:1 expansion explained above.

Regarding disk storage location, if a local disk drive is selected for storage of archived, raw data, then storage performance becomes a factor. Storage performance depends on the type of storage interface, such as SCSI-2, Fast/Wide SCSI, or RAID, and the processor input/output bandwidth. However, since SCN requirements dictate a centralized archive storage concept, wherein multiple users can access the same data at any time, local disk storage of telemetry data is not feasible.

The remaining storage alternative, namely archiving across the LAN, creates an additional LAN throughput requirement. For the worst-case configuration, wherein raw decommutated measurands are stored, the load would be a 3:1 expansion of the 128 Kbps SOH stream and the 5 Mbps mission/sensor stream, resulting in a total additional LAN throughput requirement of 15.384 Mbps. Note that the second 128 Kbps SOH stream considered in determining data rates is embedded in the 5 Mbps mission stream. A 3:1 expansion of the mission stream is also required if raw mission/sensor measurands must be processed in real-time. The more probable case, wherein raw, time-tagged mainframes are stored, results in an additional LAN throughput requirement of 5.128 Mbps, plus the small overhead for packet formatting.

When the data archiving load is included, the throughput equation becomes:

Nominal Throughputmax = 3M + 9(S1 + S2) + (M + S1) = 4M + 10S1 + 9S2

Non-Nominal Throughputmax = 3M + 18(S1 + S2) + (M + S1) = 4M + 19S1 + 18S2

where: M = Mission/sensor telemetry rate

S1 = SOH telemetry rate

S2 = Embedded SOH telemetry rate, e.g. in mission stream

Table B-1 provides values of Throughputmax (with archiving) for various telemetry rates that exist within the SCN environment.

Table B-1. LAN Throughput Requirement as a Function of Telemetry Rates (with Archiving)

S1

(Kbps)

S2

(Kbps)

M

(Kbps)

Nominal Throughputmax

(Kbps)

Non-Nominal Throughputmax

(Kbps)

0.075
---
---
7.5
1.4
1.0
---
---
10.0
19.0
8.0
---
---
80.0
152.0
64.0
---
---
640.0
1,216.0
128.0
128.0
1,000
6,432.0
8,736.0
128.0
128.0
5,000
22,432.0
24,736.0

Processing Load. The analysis of Telemetry processing load presented in Section B-6 includes estimates derived both empirically and through benchmark testing. All estimates are converted into common units of SPECint92 (see section 5.1.1 for an explanation of the performance metrics used in this report). Where performance is first calculated in units of million instructions per second (MIPS), the conversion factor to units of SPECint92 is 1/1.83.