ECSS Communications — Error Rates (space-systems/ecss/e50-error-rates)
Use when a ground network path carries a stated error-rate requirement per
ECSS-E-ST-50C clause 5.8.4, which asks the ground figure to sit well under
the rates carried by the space link and by the space network — turning those
into a budget the segments of the path can be held to, and turning a
measured path back into a verdict against them.
Domain quick reference
- An error rate without its unit is not a requirement. A rate per bit and
a rate per frame are different numbers by four orders of magnitude on a
normal transfer frame, and a document that quotes one and a test report
that measures the other agree only by accident.
- The conversion between them is not multiplication. The probability a
block carries an error is one minus the probability every bit survives,
and the linear estimate a reviewer reaches for is an upper bound that
stops being usable as soon as the product of rate and length approaches
one.
- A path is a chain, and the chain composes the same way. End to end is
one minus the product of the per-segment survival probabilities, which
for small rates is close to the sum and for large ones is a long way
under it.
- The useful inverse is apportionment. A network is procured segment by
segment, so the number each supplier can be held to is the per-segment
rate that composes to the end-to-end target, not the target itself.
- Headroom belongs in the report next to the verdict. A path that meets
its requirement with nothing to spare passes today and fails on the
first degraded hop, and the factor it holds says which of those it is.
- Volume figures are what an operator compares with a log. Errored bits
and errored seconds over a pass make an abstract rate checkable against
what the station actually recorded.
Workflow
- State the requirement with its unit and the granularity it is written
at, and state the measured or predicted rate of every segment on the
path in the same unit.
- Compose the segments into one end-to-end rate. Do not sum them; the
sum is an upper bound that overstates a lossy path.
- Convert to the granularity the requirement is written at before
comparing. A frame-level requirement against a bit-level measurement
is not a comparison.
- Grade against the requirement with a relative tolerance, and grade
again against the requirement divided by the headroom factor the
project asks for, so a path that only just passes is reported as
only just passing rather than as compliant.
- Grade the path against the clause's own comparison as well: the
composed ground-network rate beside the rate the space link carries
and the rate the space network carries, all three at one granularity.
The clause asks the ground figure to sit well under both and does not
say by how much, so state the factor the project treats as clearly
lower and report the ratio to each.
- Where the path does not reach the target, compute the per-segment
allocation that would, and name the worst segment so the effort goes
where the budget is being spent.
- Report the headroom as a ratio and in decibels, and report it as
undefined rather than infinite where a path shows no errors at all.
- Add the volume figures for the pass — errored bits, errored blocks,
errored seconds — so the prediction can be reconciled against the
station log.
Obligations
| Item |
Step |
| ECSS-E-ST-50C Rev.2 5.8.4a |
5 |
Pitfalls
- Multiplying a bit error rate by a frame length and calling it the frame
error rate. It is a bound, not the value, and it can exceed one.
- Summing segment error rates. The sum overstates the path, and for a
chain of degraded hops it can exceed one and stop being a probability
at all.
- Apportioning an end-to-end target by dividing it by the segment count.
Equal division is the right idea in the wrong algebra; the composition
is multiplicative in the survival probability.
- Reporting an error-free measurement as infinite headroom. It means the
observation window was too short to see an error, and a finite margin
figure there is an invention.
- Deciding compliance with a bare inequality against the requirement. A
path sized to land exactly on its bound is decided by rounding, and two
build hosts can disagree.
- Quoting a rate with no observation window. A rate measured over one
pass and a rate measured over a month are different claims.
Behavior contract (gate 3)
Probability, bit-count and headroom-factor validation, the block-rate and
chain compositions, the apportionment inverse and its round trip, the
undefined headroom of an error-free path, the verdict bands at the exact
requirement and at the exact headroom bound, and the per-pass volume
figures are exercised by the gate 3 contract test:
scripts/test_e50_error_rates.py against
scripts/e50_error_rates_logic.py (stdlib unittest, offline).
Run:
python3 scripts/test_e50_error_rates.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: e50-error-rates3description: "Compute the error-rate budget of a ground network data path against ECSS-E-ST-50C clause 5.8.4, which asks the ground figure to sit well under the rates carried by the space link and by the space network rather than under an absolute number of its own. Convert a bit error rate into a block or frame error rate, compose the segments of a path into one end-to-end figure, and invert that composition into the rate each hop has to hold. Report errored bits, errored blocks and errored seconds over a pass, and the headroom as a ratio and in decibels. Use when apportioning an error budget across ground segments, or grading a measured path against one. Trigger: ecss, e-st-50-ground-network, ground-network-error-rate, bit-error-rate-to-frame-error-rate, end-to-end-error-budget-apportionment, errored-seconds-per-pass, ground-segment-error-margin."4license: Apache-2.05compliance: STANDARDS-REF6standards:7 - id: ecss8 reference-only: true9gated: false10domain: space-systems11pack: space-systems12compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"13clauses:14 - standard: ECSS-E-ST-50C Rev.215 clause: 5.8.416 items: [a]17 relation: verifies18metadata:19 source: https://skillmd.com/skills/ashfordeou/e50-error-rates20 domain: space-systems21 subdomain: ecss22 tags: [ecss, e-st-50-ground-network, e50-error-rates, ground-network-error-rate, bit-error-rate-to-frame-error-rate, end-to-end-error-budget-apportionment, errored-seconds-per-pass, ground-segment-error-margin]23 version: 0.1.024 author: Aero Agent Skills25---2627# ECSS Communications — Error Rates (space-systems/ecss/e50-error-rates)2829Use when a ground network path carries a stated error-rate requirement per30ECSS-E-ST-50C clause 5.8.4, which asks the ground figure to sit well under31the rates carried by the space link and by the space network — turning those32into a budget the segments of the path can be held to, and turning a33measured path back into a verdict against them.3435## Domain quick reference3637- An error rate without its unit is not a requirement. A rate per bit and38 a rate per frame are different numbers by four orders of magnitude on a39 normal transfer frame, and a document that quotes one and a test report40 that measures the other agree only by accident.41- The conversion between them is not multiplication. The probability a42 block carries an error is one minus the probability every bit survives,43 and the linear estimate a reviewer reaches for is an upper bound that44 stops being usable as soon as the product of rate and length approaches45 one.46- A path is a chain, and the chain composes the same way. End to end is47 one minus the product of the per-segment survival probabilities, which48 for small rates is close to the sum and for large ones is a long way49 under it.50- The useful inverse is apportionment. A network is procured segment by51 segment, so the number each supplier can be held to is the per-segment52 rate that composes to the end-to-end target, not the target itself.53- Headroom belongs in the report next to the verdict. A path that meets54 its requirement with nothing to spare passes today and fails on the55 first degraded hop, and the factor it holds says which of those it is.56- Volume figures are what an operator compares with a log. Errored bits57 and errored seconds over a pass make an abstract rate checkable against58 what the station actually recorded.5960## Workflow61621. State the requirement with its unit and the granularity it is written63 at, and state the measured or predicted rate of every segment on the64 path in the same unit.652. Compose the segments into one end-to-end rate. Do not sum them; the66 sum is an upper bound that overstates a lossy path.673. Convert to the granularity the requirement is written at before68 comparing. A frame-level requirement against a bit-level measurement69 is not a comparison.704. Grade against the requirement with a relative tolerance, and grade71 again against the requirement divided by the headroom factor the72 project asks for, so a path that only just passes is reported as73 only just passing rather than as compliant.745. Grade the path against the clause's own comparison as well: the75 composed ground-network rate beside the rate the space link carries76 and the rate the space network carries, all three at one granularity.77 The clause asks the ground figure to sit well under both and does not78 say by how much, so state the factor the project treats as clearly79 lower and report the ratio to each.806. Where the path does not reach the target, compute the per-segment81 allocation that would, and name the worst segment so the effort goes82 where the budget is being spent.837. Report the headroom as a ratio and in decibels, and report it as84 undefined rather than infinite where a path shows no errors at all.858. Add the volume figures for the pass — errored bits, errored blocks,86 errored seconds — so the prediction can be reconciled against the87 station log.8889## Obligations9091| Item | Step |92|---|---|93| ECSS-E-ST-50C Rev.2 5.8.4a | 5 |9495## Pitfalls9697- Multiplying a bit error rate by a frame length and calling it the frame98 error rate. It is a bound, not the value, and it can exceed one.99- Summing segment error rates. The sum overstates the path, and for a100 chain of degraded hops it can exceed one and stop being a probability101 at all.102- Apportioning an end-to-end target by dividing it by the segment count.103 Equal division is the right idea in the wrong algebra; the composition104 is multiplicative in the survival probability.105- Reporting an error-free measurement as infinite headroom. It means the106 observation window was too short to see an error, and a finite margin107 figure there is an invention.108- Deciding compliance with a bare inequality against the requirement. A109 path sized to land exactly on its bound is decided by rounding, and two110 build hosts can disagree.111- Quoting a rate with no observation window. A rate measured over one112 pass and a rate measured over a month are different claims.113114## Behavior contract (gate 3)115116Probability, bit-count and headroom-factor validation, the block-rate and117chain compositions, the apportionment inverse and its round trip, the118undefined headroom of an error-free path, the verdict bands at the exact119requirement and at the exact headroom bound, and the per-pass volume120figures are exercised by the gate 3 contract test:121scripts/test_e50_error_rates.py against122scripts/e50_error_rates_logic.py (stdlib unittest, offline).123Run:124python3 scripts/test_e50_error_rates.py125126## Compliance127128- ECSS standards are freely downloadable (ESA); cite the source and129 paraphrase per standards-map.yaml.130- compliance: STANDARDS-REF, gated: false.