GNSS acquisition sensitivity describes how weak a signal can be when a receiver must find it under stated conditions. Tracking sensitivity describes how weak it can become while an already-acquired signal is followed. The two numbers answer different questions. Compare them only with their signal, initial information, observation time and success criterion attached; a lower tracking threshold does not establish the same cold-start capability.
For an integrator, this distinction explains a familiar result: equipment can continue operating as reception weakens, yet fail to start in that same weak-signal condition. It also explains why a sensitivity figure in dBm is not an anti-jamming rating.
Finding a signal is different from following it
Acquisition searches possible code delays and carrier-frequency offsets for evidence of a satellite signal. After detection, tracking uses the estimated alignment as a starting point and updates it as the signal changes. A receiver that already knows approximately where the signal is has a different task from one searching a much larger uncertainty region.
Lawrence Weill's original acquisition-and-tracking explanation develops this distinction for a simplified legacy GPS L1 C/A channel. It explains why tracking can work at a lower signal power than acquisition for a practical processing budget. Its example is not a fixed sensitivity gap applicable to every signal or receiver.
Tracking is not free of limits. Its estimates still have to follow the signal, and loss of synchronization can require a new search. For the underlying weak-signal extraction mechanism, the correlation guide explains why accumulating a matching code helps, and why integration time is not unlimited. The question here is how a declared threshold should be interpreted, not how to redesign that processing.
A threshold needs an endpoint and a time limit
“Acquired” might mean that a channel detected a particular satellite signal. A specification can instead describe successful receiver startup or a valid navigation fix. Those endpoints are not interchangeable: one detected signal is not a complete position solution.
The GNSS-SDR project's availability measurement guidance explicitly distinguishes single-signal tracking tests from multi-satellite tests based on retaining a position fix. It also associates acquisition with a detection probability. Read the actual procedure rather than treating every use of “sensitivity” as the same test.
The observation interval changes the claim too. Detecting within an agreed deadline is different from eventually detecting after an unspecified wait. For tracking, a momentary lock indication is different from maintaining the stated error bound over the specified interval. A comparable report identifies the endpoint, allowed time and treatment of unsuccessful trials alongside input power.
Detection probability is part of acquisition performance
A detection threshold balances missed signals against false detections. Lowering a decision threshold is not automatically an improvement if it also accepts noise peaks more often. GNSS-SDR's acquisition documentation describes detection probability and false-alarm probability together, along with code-delay and Doppler search, integration intervals and repeated search dwells.
That software implementation illustrates why “minimum input power” is incomplete without a decision rule. Two runs at the same dBm can differ in search extent, accumulated observations and the reliability required before declaring success. A successful run is useful evidence, but it does not establish the probability of success over repeated starts.
When comparing reports, retain failures as well as successes. For example, 18 successful starts in 20 attempts is a hypothetical observed fraction of 90%, not a demonstrated universal 90% success probability. It also says nothing about a different input power, deadline or startup state. This is an arithmetic example, not a recommended sample size or an acceptance threshold.
Keep four specifications separate
| Specification | What it is describing | Comparison that would be misleading |
|---|---|---|
| Acquisition sensitivity | A minimum wanted-signal level for the specified acquisition task and initial state | Substituting a tracking result for startup from unknown synchronization |
| Tracking sensitivity | A minimum level for continuing the defined tracking or navigation condition | Assuming that the receiver can first acquire at that same level |
| Time to first fix | Elapsed time to a defined first valid navigation result under stated inputs | Applying a TTFF measured at a stronger signal to the sensitivity threshold |
| Reacquisition | Recovery of the defined signal or navigation condition after a specified interruption | Treating retained information and the interruption history as irrelevant |
These are different axes, not four competing grades. A project may need both a startup threshold and an ongoing-operation threshold. The existing startup, tracking and recovery evaluation section addresses the corresponding event sequences; it does not make their numerical thresholds equivalent.
Initial information changes the startup task
Cold, warm and hot starts describe different available information, not different jammer powers. In its archived u-blox 8/M8 receiver description, section 14, u-blox distinguishes a broad search without prior information from starts retaining approximate time, position, almanac or valid ephemeris. The same section warns that its reset command does not clear assistance data held in flash. Those details belong to that receiver family, not to all implementations.
For a comparison, describe the information actually available at the start, not just the reset button pressed. An assisted start with valid external data and an unaided start must not quietly become one “acquisition sensitivity” result. If assistance is part of the intended deployment, assess that intended configuration rather than automatically discarding it; simply keep its dependency visible.
Read the row and its footnotes together
A concrete manufacturer example is the u-blox MAX-M10S datasheet, UBX-20035208 R08, page 5. Its multi-constellation table separately lists typical tracking/navigation, reacquisition, cold-start and hot-start sensitivity. The values are respectively −167, −160, −148 and −159 dBm for the configurations shown. The sensitivity footnote says that these figures were demonstrated with a good external LNA and measured at room temperature.
The same page gives TTFF under a separate condition: all satellites at −130 dBm, with its own footnotes. Therefore, the listed cold-start TTFF is not a promise to obtain that fix in the same time at the cold-start sensitivity limit. The document also gives different single-GNSS entries. Keep the relevant table, mode and footnotes together instead of lifting the lowest number from the page.
These are attributed u-blox product facts, not ResiNav specifications or evidence of compatibility. Their value here is the reading method: separate the operating state, power threshold, time result and RF setup before comparing products.
What a six-decibel difference does—and does not—mean
Consider an invented, same-reference-point example. A receiver's defined acquisition test passes down to −145 dBm, while its separately defined tracking test passes down to −151 dBm. Assume the same wanted signal and RF configuration, and keep each test's endpoint and duration explicit.
The tracking threshold is 6 dB lower in wanted-signal power. Using the power-ratio convention explained by NIST's decibel reference, section 8.7, the corresponding ratio is:
10^((−145 − (−151))/10) ≈ 3.98.
Thus the acquisition threshold uses about four times the wanted-signal power of the tracking threshold in this example. A signal at −149 dBm lies below the stated acquisition threshold but above the tracking threshold. It is consistent with those declared limits for an already-tracking receiver to continue while a new start does not meet the acquisition test.
The 6 dB is not measured jammer suppression, an extra 6 dB of system J/S tolerance, or four times better positioning accuracy. No jammer level was even specified. It is a difference between two wanted-signal thresholds for two different tasks, not a margin that can be added to an anti-jamming product figure.
Match the RF reference before ranking receivers
Signal power must belong to an identified point in the tested path. A generator setting before a cable or attenuator is not automatically the power delivered at the equipment input. An external amplifier also changes the setup; neither its presence nor its gain can be silently dropped from a comparison.
The u-blox production-test application note, section 3 describes a repeatable simulator-based sensitivity check and a reference-device comparison. Its procedure includes controlling additional gain and the applied level. That is a particular production-test method, not a universal GNSS qualification standard or a service offered by ResiNav.
This is why a bare module figure and an assembled-device result may not have the same boundary. Likewise, sensitivity in dBm and a receiver's reported C/N0 in dB-Hz are different quantities. The C/N0 and SNR guide explains the units; do not convert between them without the noise reference and bandwidth conventions needed for that conversion.
Choose the number that answers the operating question
For equipment that must switch on at a weak-signal location, compare acquisition evidence for its actual retained information and deadline. A unit that starts elsewhere and later enters attenuation poses a different question: compare the defined tracking/navigation limit and the required continuity. If the signal is interrupted, use the corresponding reacquisition condition rather than extending either result by assumption.
Weak wanted signals and added interference are also different input conditions. A low sensitivity threshold does not establish behavior under a specified interferer, front-end overload or an authenticity threat. Use the anti-jamming datasheet review for that separate metric comparison. When a project independently needs reception protection, the ResiNav catalogue is the model-screening route; no acquisition or tracking sensitivity is assigned to a ResiNav model by this article.
The practical result is a fairer comparison: the right signal, at the right reference point, in the right initial state, achieving the stated result within the stated time. A more negative number becomes meaningful only after those parts of the claim are aligned.
NEXT STEP
Turn available platform information into an engineering review.
Use the Technology centre to frame the discussion, review relevant application scenarios, then send the available platform and receiver details for confirmation.