A short packet can fit inside a slowly varying channel, while a long packet can experience significant gain and phase change before it ends. The comparison depends on motion, carrier band, antennas, correlation threshold and the exact packet interval.
This guide builds a temporal-correlation ledger, interpolates one fictional threshold crossing, and compares short, boundary-sensitive and long packets. It preserves uncertainty and receiver pilot cadence.
NIST channel measurements provide the research basis. Worked numbers are stipulated teaching examples and do not predict a specific deployed radio link.
In the central record, normalized temporal correlation is 0.82 at 8 ms and 0.44 at 16 ms. Linear interpolation for a 0.5 rule gives about 14.7 ms. Packets of 2 ms and 30 ms therefore occupy about 0.14 and 2.04 coherence-time units.
This specialist guide belongs to the Wireless Communication root. It also connects to The Gold Standard of Education | Learning That Transfers into Life and the How X Works hub. The worked cases are original teaching examples; their numbers are stipulated so that the reasoning can be checked.
eduKateSG · Wireless Communication
Find your next learning step
Coherence time summarizes how quickly a propagation channel decorrelates in time under a stated rule. Packet duration describes how long a particular transmission occupies the air. Their ratio guides a model; it does not make them the same quantity.
ROUTE 5 · CHAPTERS 21–28
Report and continue
Add receiver checks, build a receipt, repair claims and use NIST sources.
Full chapter contents · Wireless Communication subject guide
Chapter contents
Define the clocks · Chapters 1–5
2. Define packet duration precisely
3. Define coherence time by a threshold
Connect motion · Chapters 6–10
7. Separate Doppler shift from coherence time
8. Avoid one reciprocal constant
Work receiver timing · Chapters 11–15
11. Match channel estimation cadence
12. Distinguish coherence time from latency
Audit the evidence · Chapters 16–20
16. Check sounder snapshot time
17. Put uncertainty around the crossing
Report and continue · Chapters 21–28
21. Check variation within one OFDM symbol
22. Separate pilot spacing from packet length
23. Treat antenna links as a matrix of clocks
24. Avoid turning adaptation into measurement
25. Build a channel-timing receipt
Start with the mechanism, not the symptom. Coherence time summarizes temporal correlation of a changing propagation channel, while packet duration is a property of a transmitted frame or protocol unit. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: A 2 ms packet crosses a channel estimated to remain strongly correlated for 20 ms under a stated threshold. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Preserve the correlation definition, carrier band, motion state and packet timing convention. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Keep the channel fixed and extend the packet to 30 ms. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. Packet duration may include preamble, payload, pilots, guard intervals and retransmission gaps depending on the question. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: A 1 ms payload sits inside a 1.4 ms physical frame; using the shorter value hides channel changes during training and guards. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Mark start and end at the layer whose receiver processing is being assessed. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Compare one physical frame with an application message split across ten frames. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. Coherence time depends on a temporal autocorrelation rule and threshold rather than being a universal clock reading. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: Correlation falls from 0.82 at 8 ms to 0.44 at 16 ms; a 0.5 threshold brackets the crossing. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Report the bracket and interpolation assumption instead of attaching false precision. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Change the threshold to 0.7 and recompute. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
A useful explanation preserves cause, observation and decision. Linear teaching interpolation between 8 ms at 0.82 and 16 ms at 0.44 places the 0.5 crossing about 14.7 ms. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: The needed drop is 0.32 of a total 0.38 across 8 ms, giving 8+0.32/0.38×8. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Retain the raw samples because the true correlation curve need not be linear. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Insert a measured point at 12 ms with correlation 0.60. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Start with the mechanism, not the symptom. A 2 ms packet has ratio Tpkt/Tc about 0.14 using the 14.7 ms teaching estimate. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: The channel may be treated as approximately constant during that frame for a first model, subject to direct evidence. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Inspect channel estimates across preamble and payload rather than converting the ratio into a guarantee. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Increase speed while leaving packet timing fixed. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. A 30 ms packet has ratio about 2.04 and can span multiple coherence intervals. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: One channel estimate at the packet start may age before later symbols arrive. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Measure error, phase and gain evolution across the complete frame and receiver update schedule. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Divide the transmission into shorter independently trained frames. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. Doppler shifts of multipath components contribute to Doppler spread, while coherence time summarizes resulting temporal correlation under a convention. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: A single maximum-speed formula may not describe a rich industrial channel with several moving scatterers. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Use a measured Doppler spectrum or justified model and name the conversion convention. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Narrow the antenna beam and remeasure admitted multipath. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
A useful explanation preserves cause, observation and decision. Rules such as 1/fd or 0.423/fd encode different definitions and channel assumptions. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: At maximum Doppler 100 Hz they give 10 ms and 4.23 ms, a large design difference. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Never mix a constant from one correlation criterion with a threshold from another. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Use the measured autocorrelation crossing instead of a rule of thumb. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Start with the mechanism, not the symptom. Maximum Doppler scales with relative speed and carrier frequency under the simple projection model. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: A moving receiver at the same speed can see larger Doppler at 28 GHz than at 3.5 GHz. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Record velocity vector, wavelength, antenna pattern and scatterer motion. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Keep speed fixed while changing the carrier band. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. An ensemble coherence time can hide a rapid local change within one packet. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: A measured frame loses 20 dB over 3 ms even though the broader average crossing is 15 ms. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Inspect per-frame channel trajectories and uncertainty, not only the pooled summary. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Stratify the route by motion and environment. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. Receiver pilots and tracking loops determine how often channel knowledge is refreshed. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: A 20 ms packet carries pilots every 2 ms, so it need not rely on one start-of-frame estimate. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Measure estimator error and interpolation under the actual pilot pattern. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Remove every second pilot and compare held-out performance. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
CHAPTER 12 OF 28 · Work receiver timing
12. Distinguish coherence time from latency
A useful explanation preserves cause, observation and decision. Coherence time describes temporal similarity of the propagation response; latency measures delivery timing through a system. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: A packet can arrive late over a slowly varying channel because of queues, scheduling or retransmissions. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Use separate channel and network traces with synchronized clocks. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Hold the channel static and increase queue occupancy. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
CHAPTER 13 OF 28 · Work receiver timing
13. Distinguish packet duration from interarrival time
Start with the mechanism, not the symptom. A short packet may be followed by a long idle gap, while burst packets can overlap one coherence interval differently. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: Ten 1 ms packets sent every 50 ms do not form one 10 ms continuous channel exposure. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Preserve on-air intervals and gaps in the timeline. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Transmit the same packets back-to-back. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. A retry can occur in a new channel realization even when each individual packet is shorter than coherence time. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: The first frame fails at time zero and a retry starts 25 ms later with Tc near 15 ms. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Treat attempts separately and record scheduler delay, combining and channel estimates. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Retry after 5 ms instead. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. Subcarriers may be frequency-flat while the channel still varies over the duration of a long multicarrier frame. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: Coherence bandwidth supports the frequency model and coherence time supports the temporal model; both axes matter. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Inspect the time-frequency channel surface and pilot lattice. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Widen the signal but shorten the frame. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
A useful explanation preserves cause, observation and decision. A channel sounder must acquire a snapshot quickly enough that the environment is effectively stable during that measurement. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: A 12 ms switched-array scan can smear a channel whose coherence time is near 8 ms. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Record scan order, shutter time and motion; a blurred estimate cannot certify packet-scale stationarity. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Use a faster scan or fewer antenna states. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Start with the mechanism, not the symptom. Finite records and nonstationary motion make temporal-correlation estimates variable. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: Resampling yields a 0.5-crossing interval from 12 to 18 ms rather than one exact 14.7 ms value. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Classify packets against the whole interval and identify boundary-sensitive cases. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Test packets of 2, 15 and 30 ms. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. Tpkt/Tc is a screening ratio, not a universal threshold separating slow from fast fading. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: Ratios 0.1, 0.9 and 3.0 suggest different modelling risks but do not calculate error rate. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Validate the receiver metric that matters under representative records. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Change modulation and coding while keeping the channel trace. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. A channel can retain amplitude correlation while phase rotates enough to challenge coherent demodulation. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: A packet shows modest gain change but accumulating phase error across its payload. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: State whether temporal correlation is complex, magnitude-only or power-based. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Compare coherent and differential receiver assumptions. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
A useful explanation preserves cause, observation and decision. Coherence time can vary by route segment, machinery state, people movement and antenna orientation. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: A factory aisle and an open loading bay produce different Doppler spectra in the NIST campaign. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Report distributions and conditions rather than one building label. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Use a conservative percentile for a defined deployment population. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Start with the mechanism, not the symptom. When the channel changes materially during a multicarrier symbol, loss of orthogonality and inter-carrier interference can matter even if the whole packet comparison looks acceptable. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: A packet contains many short symbols; its total duration is below the chosen coherence time, yet the highest-mobility case makes the per-symbol channel evolution measurable. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Compare symbol duration, Doppler spread and the receiver model rather than applying one packet-level threshold to every impairment. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Shorten the symbols while keeping total packet duration and motion fixed. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
CHAPTER 22 OF 28 · Report and continue
22. Separate pilot spacing from packet length
Keep the two objects in separate ledger columns. Pilot observations sample the channel in time, while packet duration merely bounds the transmitted interval; two equal-length packets can support different interpolation quality. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: Packet A places pilots every millisecond and Packet B only at its start, although both last eight milliseconds. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Reconstruct the assumed channel between pilots and inspect residual error under the same temporal-correlation trace. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Keep packet length fixed and halve the pilot spacing, then keep spacing fixed and double the packet. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
CHAPTER 23 OF 28 · Report and continue
23. Treat antenna links as a matrix of clocks
The short answer becomes reliable only when its boundary is named. A multiple-antenna channel contains several link coefficients whose temporal correlations need not cross a threshold at the same instant. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: One transmit-receive pair decorrelates near twelve milliseconds while a second remains above the threshold at twenty milliseconds. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Report the aggregation rule—minimum, percentile, mode or link-specific value—before comparing the channel with a packet. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Change the beam or antenna pair without changing packet timing. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
CHAPTER 24 OF 28 · Report and continue
24. Avoid turning adaptation into measurement
A useful explanation preserves cause, observation and decision. Shortening packets after errors rise can improve robustness, but the adaptation decision does not itself estimate coherence time or identify Doppler spread. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: A scheduler halves frame duration when retransmissions increase during motion; interference and congestion remain alternative causes. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Require channel estimates, motion evidence or a controlled model before assigning the improvement to temporal coherence. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Hold motion fixed, vary interference, and observe whether the same adaptation triggers. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Start with the mechanism, not the symptom. A reproducible record includes band, antennas, location, speed, correlation statistic, threshold, observation window, packet boundaries and pilot cadence. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Worked case: Two teams can then reproduce the ratio and explain different results. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Evidence check: Store sampled correlation values and unresolved crossings with the headline estimate. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Changed-context transfer: Move the same radio to a new route and repeat the receipt. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Keep the two objects in separate ledger columns. A packet shorter than coherence time may support a quasi-static model, but it is not guaranteed error-free or unchanged. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Concrete case: Noise, interference, deep fading, estimation error and protocol behaviour remain separate causes. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Practical check: Rewrite conclusions with the stated threshold and evidence domain. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Transfer test: Ask what observation would falsify the quasi-static assumption. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
The short answer becomes reliable only when its boundary is named. NIST measurements analyse Doppler spectra, Doppler bandwidth and coherence time across industrial environments and conditions. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Reason through this case: Read https://www.nist.gov/publications/empirical-characterization-doppler-industrial-wireless-channels and the NIST dynamic-channel study at https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=904976. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Audit the claim with this evidence: Check paper definitions before importing a reciprocal rule or threshold. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Now change one condition: Recompute the teaching cases under the paper’s selected convention. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
A useful explanation preserves cause, observation and decision. The coherence-bandwidth guide owns frequency correlation; this guide owns temporal correlation against packet duration. State which property is being changed, which box, channel, credential or measurement owns it, and which conclusion is still undecided.
Use this stipulated example: Continue with https://edukatesg.com/portfolio/the-gold-standard-of-wireless-communication-separating-coherence-bandwidth-from-signal-bandwidth/ and the packet-delay-variation sibling. Keep every stipulated value and comparison visible so another reader can reproduce the result without supplying a hidden default.
Verify rather than assume: Return to the Wireless Communication root for channel width, throughput, emissions and clock drift. Record the definition, version, measurement window or processing rule that makes the evidence relevant, together with any unresolved condition.
Test whether the reasoning travels: Choose the route by the axis and object of the reader’s question. Re-run the decision from the changed input instead of copying the earlier label. Explain what changed, what stayed fixed and what the example cannot prove.
Contents · Previous chapter · Return to Wireless Communication
Continue through the verified reading route
Coherence bandwidth and signal bandwidth
Coherence bandwidth and signal bandwidth. Use this for variation across frequency rather than variation during a packet.
Wrapped phase and delay ambiguity
Wrapped phase and delay ambiguity. Use this when a repeating phase measurement leaves more than one possible delay.
Retry time and unique-payload goodput
Retry time and unique-payload goodput. Use this when retransmission airtime changes delivered useful data per unit time.
Frequently asked questions
Is coherence time a fixed property of a site?
No. It depends on the link, carrier band, motion, antennas, environment and the correlation definition used.
Must a packet be shorter than coherence time?
That comparison is a model choice, not a universal pass mark. Receiver estimation, pilots, modulation and acceptable error also matter.
Is packet duration the same as latency?
No. Packet duration is an on-air interval under a stated boundary; latency can include waiting, processing, retransmission and routing.
Can retransmission results reveal coherence time?
They can motivate investigation, but congestion, interference and protocol choices can produce similar symptoms.
Authoritative sources and further reading
NIST: Empirical Characterization of Doppler in Industrial Wireless Channels. Measurement context for Doppler spectra and temporal channel variation.
NIST publication PDF. Full technical report supporting the measurement distinctions used in this guide.
Return to the Wireless Communication subject guide, The Gold Standard of Education, or the How X Works hub.