Cellular Radio for Product Engineers

01 / 10

“Global LTE” is not a product requirement

The module says “global LTE” and 5 µA sleep. Why can the installed node still miss every urgent command?

The label leaves out the actual network service, the host antenna path and the state in which a command arrives. A module may implement a radio mode that the final host cannot radiate effectively. A registered device may intentionally be unreachable while sleeping. None of those questions is answered by a band list.

Think about itWhich missing evidence could overturn a shortlist before comparing battery size?

Answer

Any explicit required-tuple incompatibility can stop it: module, host, network, peer/configuration or simultaneous combination. Missing evidence instead keeps it under inspection. A successful radio transfer still needs a separate application deadline test.

Illustrative remote-fleet contract · inherited from 07.4

Chosen field goal: ≥99% complete unique 32-byte D0 reports at server D0 within 60 s in each site/orientation stratum, one report/600 s. Urgent commands accompany 1% of reports; ≥99% at node D0 within 30 s of server release at arbitrary reporting phases. No measured achievement.

1000 fixed nodes; metal-machine and open-pole installations, upright and rotated enclosures. Count generated sequence IDs, unique complete D0 delivery, omissions and late reports in each stratum. These are chosen requirements, not results.

Technical shortlist = module ∧ host ∧ network ∧ peer/configuration ∧ combinationThe conjunction is over an exact requested tuple. One false predicate gives fail; no false but an unknown gives inspect; all true and current gives supported in the fixture. Qualification, operator acceptance and market applicability remain separate gates.

The first checked example is Node-A/T1 in Region-A: all five predicates are satisfied in the fictional data. Changing only the request to T2 makes the host predicate false. The module has not changed; the product claim has.

Common misconceptionA standards label is a product requirement.

A label cannot specify the D0 endpoint, loss denominator, deadline, installation, mobility or infrastructure owner. Without those, two apparently similar radios can be evaluated against different jobs.

Go deeperBring the correct boundaries forward

07.1 owns evidence reading; 07.4 freezes the remote-fleet traffic. Revisit sensitivity and RF budgets and antennas in real products. The generic 2.450 GHz QPSK portfolio example is preserved; it never becomes a cellular PHY by relabeling it.

Decision: freeze the service and reference planes first. Then find which documents can support this particular claim.

02 / 10

Navigate releases and specification versions

Two documents say Release 13. Can their version numbers be treated as interchangeable?

A TS specifies technical requirements; a TR records report material. A release groups coordinated specification work. Stages describe service requirements, system architecture and detailed protocols. A work item organizes deliverables; a study item investigates feasibility. An exact document version identifies a particular text. [WORKING]

A real document chain · navigation, not a conformance claim
Step	Pinned evidence / remaining work
RF source	3GPP TS 36.101 V13.25.0, Release 13; ETSI publication July 2023. Read the mode-specific clause, not just the cover.
Change history	June 2018, RAN #80, RP-181111, CR 5093: eMTC subband CQI correction, recorded into V13.12.0. History entry read; original CR text and its product impact remain unverified.
Capability source	TS 36.306 V13.5.0: per-band and combination fields are separate. Obtain a matching device capability record.
Conformance dependency	TS 36.521-1 V13.4.0 §4 directs requirements to the Release 14 document. The exact destination revision and applicable tests must be resolved before using a limit.

A change request records a proposed change against identified text; working-group agreement and TSG approval precede incorporation. An erratum/correction is another reason to inspect affected text and status, not an automatic hardware update. The x.y.z convention distinguishes release/approval, technical and editorial revision roles. It does not prove interoperability. [WORKING] [LTE-RF] [LTE-TEST]

Think about itWould selecting a numerically newer version make a previously unknown compatibility edge true?

Answer

No. Record an explicit applicability/compatibility edge with its reviewer and source. In the matrix, fw-v2 without matching evidence stays inspect. The named host-v2 example supplies a separate design-evidence revision; changing a requested version alone does nothing.

Common misconceptionA release identifies every deployed feature.

Implementation choices, capability signaling, network configuration and subscription can all constrain use. Keep the release as one identity field, then verify each dependency for the selected configuration.

Go deeperA newer catalogue entry does not erase a legacy mode

The consulted ETSI directory lists TS 38.101-1 V18.14.0 alongside older revisions. This lesson pins accessible V18.10.0 for scope orientation; it does not assert that it is newest or transfer its requirements to a product. Overdue review means recheck the evidence. It does not prove the old statement became false.

Decision: carry document identity, consulted locator, version edge, access date and review trigger into the card. Next select the radio family that serves the traffic.

03 / 10

Choose the service and radio family

Should a 32-byte remote report and a higher-rate gateway buy the same radio family?

Begin with the work each endpoint must perform. A node spends long intervals waiting; a gateway may sustain backhaul traffic and several RF chains. Small uplink payloads do not imply that urgent downlink, mobility, firmware updates or service continuity are easy.

Family orientation · questions to resolve, not a feature guarantee
Candidate family	Product fit to investigate	Evidence that can reverse it
LTE / E-UTRA	An established cellular radio family with different UE capability sets.	Exact category, mode, duplex, operator service and lifecycle; do not infer them from “4G”.
LTE-M	A constrained LTE node candidate; investigate periodic telemetry plus command reachability.	Granted sleep behavior, mobility/registration traces, supported bands and exact firmware.
NB-IoT	A distinct cellular-IoT candidate; examine narrowband resource use and long state timelines.	Deployment mode, scheduling, repetitions, roaming and command deadlines need specific evidence.
NR	A gateway candidate when its required capacity and network architecture justify the RF burden.	Standalone/interworking identity, relevant RF part, bands, bandwidth, MIMO and combination support.

Real implementation example: Nordic’s nRF9160 modem firmware 1.3.7 notes identify Release 13 LTE-M/NB-IoT support with selected later features and list LTE-M Band 20. The download page targets revision 2 (B1). Our example is a CH candidate LTE-M data service, evaluated 2026-09-08; the exact order code, host antenna, operator, subscription and deployed features remain unverified. [NORDIC-FW] [NORDIC-SKU]

Think about itDoes that product statement support a claim that urgent commands reach our CH node in 30 s?

Answer

No. It supports a bounded firmware capability statement. The carrier and product owners still need provisioning, reachability and end-to-end timing evidence at arbitrary sleep phases.

Useful service = unique complete D0 reports delivered by the deadline / D0 reports generatedBoth counts refer to the same observation interval and installation stratum. D0 is application data; D1 is the framed/coded representation. PHY bit rate, successful copies and useful delivery are different quantities.

Go deeperWhich RF document belongs to the gateway?

TS 38.101-1 V18.10.0 is explicitly the UE Range 1 Standalone part. “38.101” alone is insufficient. Re-select the applicable part and conformance/capability sources if the proposed gateway is FR2 or uses interworking. Base-station requirements cannot stand in for UE requirements. [NR-RF]

Decision: preserve separate LTE-M, NB-IoT and NR variants. Do not copy the node event ledger into a gateway power claim.

04 / 10

Build complete band and duplex tuples

The band, resource class and mode each appear somewhere in a module table. Does their combination exist?

A tuple is a complete configuration, not a bag of features. FDD separates uplink and downlink frequency ranges; TDD shares a band through a time allocation. Half-duplex capability still matters for a particular FDD device. Carrier bandwidth, a bandwidth part and scheduled resources identify different scopes: an NR BWP is a contiguous resource-block subset for a given numerology, not the entire carrier. [LTE-CAP] [NR-PHY]

Synthetic whole tuples · no real frequencies or UE categories
Tuple	Mode / band / duplex	Resource / power scope / RF path
T1	LTE-M-like node / TEACH-L / FDD	C1; synthetic resource; PC-L; no genuine UE class; L-path-v1. Synthetic identity v1.
T2	NB-IoT-like node / TEACH-N / FDD	C2; synthetic resource; PC-N; no genuine UE class; N-path-v1. Synthetic identity v1.
T3	NR-like gateway / TEACH-H / TDD	C3; synthetic resource; PC-H; no genuine UE class; H-path-v1. Synthetic identity v1.

Think about itNode-X supports (T1,C1) and (T2,C2). May we request T1 with C2?

Answer

No. Neither list entry contains that complete request. With a complete set, the missing cross-product fails. A partial list would leave it unknown. The counterexample preset deliberately asks for this unsupported pair.

Requested tuple ∈ module set ∩ host set; network(tuple) ∧ peer(tuple) ∧ combinationMembership compares every identity field: RAT/mode, source version, band, duplex, resource, power-class scope and RF path. The network is three-valued evidence. A simultaneous request also needs its exact combination identifier.

Frozen capability truth table · synthetic v1
Request	Technical result	Decisive evidence
Node-A / Region-A / T1	supported in fixture	All exact predicates true; actual-network, qualification and market gates remain unknown.
Node-A / Region-A / T2	fail	host:T2:absent-complete
Node-B / Region-A / T2	supported in fixture	All exact predicates true; actual-network, qualification and market gates remain unknown.
Node-B / Region-B / T2	inspect	network:T2:unknown
Node-A / Region-B / T1	fail	network:T1:explicit-false
Gateway-C / Region-B / T3	supported in fixture	All exact predicates true; actual-network, qualification and market gates remain unknown.
Gateway-C / Region-A / T3	fail	network:T3:explicit-false
Gateway-C / Both-singles / T1+T3	fail	combination:T1+T3-simultaneous-v1:explicit-false

Common misconceptionA partial list’s missing item is definitely unsupported.

Only a declared complete set permits that inference. Unknown support blocks a recommendation without inventing an incompatibility. Conversely, one explicit false defeats the conjunction even when another predicate is unknown.

Go deeperLocate a real band without granting a network

TS 36.101 V13.25.0 Table 5.5-1 defines E-UTRA Band 20 as FDD: UE transmit 832–862 MHz, UE receive 791–821 MHz. §5.5E includes Cat-M1 band applicability. That definition supplies no current country allocation, operator service or installed-host result. [LTE-RF]

Decision: require exact tuple membership and carry every false/unknown reason. Now connect that tuple to waveform and RF hardware.

05 / 10

Connect waveform to front-end power

Why can a modem’s maximum power label fail to predict either its RF waveform or its supply pulse?

Adding modulated subcarriers produces a varying envelope. A PA must preserve the selected waveform’s linearity while supplying its peaks. LTE PUSCH uses transform precoding in the SC-FDMA family; the selected NR PUSCH specification contains configured transform-precoding branches. “Uplink” alone does not identify every physical channel or waveform. [LTE-PHY] [NR-PHY]

$\begin{aligned}\mathrm{PAPR}&=\frac{\max|x(t)|^2}{\mathrm E\{|x(t)|^2\}}\\P_{\mathrm{DC,PA}}&=\frac{P_{\mathrm{RF,out}}}{\eta_{\mathrm{PA}}}\end{aligned}$ x(t) is the complex envelope over a declared observation interval; PAPR is a linear peak-to-mean power ratio, or 10 log₁₀ of it in dB. η_PA is PA drain efficiency under the stated waveform and operating point. The second relation covers the PA supply only, not the whole modem.

Think about itIf an idealized PA emits .10 W at efficiency .25, is the whole radio a .40 W load?

Answer

The PA portion is .10/.25 = .40 W under those assumptions. Other radio circuits, conversion loss and state timing remain outside that example. It is not a measured modem power value.

Real RF trace · deliberately incomplete product test specification
Field	Located evidence or explicit gap
Identity / direction	LTE-M Cat-M1 candidate; TS 36.101 V13.25.0; Band 20 FDD UE uplink; nRF9160 revision 2, firmware 1.3.7 capability statement.
Mode-specific requirement	§6.2.2E identifies Cat-M1 maximum-output-power requirements and the ≥1 subframe (1 ms) measurement interval. §6.2.3E treats resource/modulation-dependent maximum power reduction.
Conditions to complete	Exact half/full-duplex capability, channel bandwidth, resource allocation, modulation, power class, reduction conditions, firmware configuration and antenna-port mapping are unresolved. No numeric TX limit is adopted.
Conformance context	§4.1 distinguishes minimum RF requirements from test tolerances. Selected TS 36.521-1 V13.4.0 routes to Release 14; lab must pin the applicable destination tests and revision.
Plane / statistic	Conducted UE output at its defined antenna connector; map that connector to product R1/R2 and document fixture/feed loss. This is neither period-average RF output nor S0 TRP.

[LTE-RF] [LTE-TEST] [NORDIC-FW]

Common misconceptionMaximum power class is the average transmit power.

Power control, allocation, power reduction, waveform quality and time spent transmitting all matter. PA efficiency changes with operating point. Thermal load and battery droop require their own time intervals and supply measurements.

Go deeperName the quality failure before choosing backoff

Compression can increase modulation error and adjacent-channel energy. The corrective backoff depends on the actual PA, allocation and quality requirement; there is no universal “OFDM penalty”. Revisit transmit power, fidelity and efficiency for the RF-chain budget.

Decision: leave the unverified RF limit open, assign it to the standards and lab owners, and preserve the product plane transformation.

06 / 10

Antenna and receiver requirements follow the tuple

Why do two individually supported radios still fail when the gateway transmits them simultaneously?

The combination may need separate filters, duplexers, receive paths, isolation, supply headroom and antenna behavior. Strong local transmit energy can desensitize a receiver through leakage, blocking or nonlinear mixing. CA, dual connectivity and concurrent independent radios describe different configurations; a list of single bands does not establish any of them. The capability specification has a separate combination field. [LTE-CAP]

Think about itBoth-singles network evidence supports T1 and T3. Is Gateway-C’s simultaneous request supported?

Answer

No. The exact combination predicate is false even while both single-mode predicates are true. Keep that decisive reason independent of the network records. Do not create a virtual extra RF path by intersecting band names.

Conceptual plane sequence, with intervening stages abbreviated. R3 is the receiver decision boundary, never its antenna input. D1 identifies framed/coded data between the PHY/MAC and application. Feed, mismatch and radiation losses each enter once; SI and e^+jωt convention.

$\Delta G_{\mathrm{realized,dB}} = 10 \log _{10}(\frac{\eta _{\mathrm{rad,2}}}{\eta _{\mathrm{rad,1}}}) = -3.01029995664 \mathrm{dB}$ This controlled comparison changes radiation efficiency from .50 to .25 at fixed directivity in the stated direction, match, feed power and propagation. Efficiency is dimensionless. The differential is derived; no additional range, repetitions, transmit power or search time is predicted.

Receiver and antenna evidence follows the tuple
Quantity	Required context / consequence
Conducted sensitivity	R1 or mapped R2 input, exact mode/release, bandwidth/resources, wanted waveform, propagation model, test statistic and throughput/error criterion. No generic receiver threshold is used here.
Blocking / desensitization	Wanted level, interferer frequency and power, simultaneous path, filter state and acceptance statistic. Include host clocks and other radios.
TIS / TRP	S0 radiated receiver/transmitter metrics under their specified spatial procedures; not a conducted threshold or a service-availability percentage.
Gain / loss accounting	Ask whether quoted gain already includes radiation efficiency, mismatch and feed loss. Add only missing transformations, once.

Common misconceptionPoor antenna efficiency directly multiplies modem DC energy.

The efficiency differential changes the RF budget. Network adaptation may then alter operating states, but that needs evidence. Multiplying current by 1/η invents a control policy and confuses radiation efficiency with PA efficiency.

Go deeperOne mounted antenna is a distribution, not one gain number

Enclosure, machine surface, cable routing and orientation affect pattern and match. The antenna owner should report mounted distributions and the RF paths available in each combination. Revisit channel and link distributions; no fixed-radius coverage circle belongs in this decision.

Decision: require an antenna/front-end plan for the chosen tuple, then examine how the radio spends time at the supply boundary.

07 / 10

Integrate the whole network-state timeline

Can a rare search cost more charge than hundreds of seconds of low-current sleep?

A device can search, register/attach, resume service, exchange scheduled payload, remain in a connected or idle tail, and sleep. A registered report does not necessarily repeat attach. Mobility and lost service can introduce reselection, tracking and new search events. Requested PSM/eDRX timers do not prove a grant or a measured current state. [LTE-STATE]

S32-v1 registered service · synthetic assumptions, not measurements
State	Duration (s)	Current (mA)	Charge (mC)
Resume / wake	0.04	50	2
Connected payload transfer	0.2	80	16
Connected / idle tail (once)	2	5	10
Residual PSM-like sleep	597.76	0.005	2.9888

$Q = \sum I_{i}t_{i}\qquad E = VQ\qquad I_{\mathrm{avg}} = \frac{Q}{T}\qquad t_{\mathrm{sleep}} = T - \sum t_{\mathrm{active}}$ Iᵢ is total radio-supply current in amperes during serial state i; tᵢ and reporting period T are seconds. Q is coulombs, constant V is volts, E joules and I_avg amperes over the entire period. Here T = 600 s and V = 3.6 V. No board, sensor, regulator or battery-aging model is included.

The registered charge is 50×.040 + 80×.200 + 5×2.000 + .005×597.760 = 30.9888 mC using mA and seconds. At 3.6 V that is 111.55968 mJ and 51.648 µA period average. PSM-only 5 µA misses the event charge.

Think about itInsert 30 s of successful search at 40 mA. Does that merely add one more small state?

Answer

It displaces 30 s of .005 mA sleep: ΔQ = (40−.005)×30 = 1199.85 mC. Total Q becomes 1230.8388 mC. Search success in 30 s is stipulated in this branch; a real network may time out instead.

Illustrative serial states at 3.6 V and 600 s. The outlined abort bar has zero delivered payload; a charge bar is never a delivery-success indicator. No measured distribution is implied.

Independent numerical anchors · 32 D0 bytes / 600 s / 3.6 V
Scenario	Q (mC)	E (mJ)	I average (µA)
Registered	30.9888	111.55968	51.648
Search succeeds	1230.8388	4431.01968	2051.398
Mixture s = .1	150.9738	543.50568	251.623
Search aborts	1204.8498	4337.45928	2008.083

$Q_{\mathrm{mix}} = (1-s)Q_{\mathrm{registered}} + s Q_{\mathrm{search}}\qquad s = .1 \to Q_{\mathrm{mix}} = 150.9738 \mathrm{mC}$ s is an illustrative independent fraction of generated-report events that search successfully. The weighted mean is 543.50568 mJ and 251.623 µA at the stated V and T. It is not measured coverage, P95 or worst-case behavior. Unknown s leaves two scenarios, not an estimated mean.

Stop failed search at the correct branch

The abort policy wakes for .04 s, searches for 30 s without registration, sends zero payload and sleeps/backoffs for 569.96 s. It has no connected tail. Charge is 1204.8498 mC, energy 4337.45928 mJ and average current 2008.083 µA. Energy per delivered report and delivery latency are unavailable because nothing is delivered.

$\begin{aligned}K&=r+1\\A&=\sum_{k=0}^{K-1}p^k\\q&=1-p^K\end{aligned}$ Separate retry fixture: every connected attempt spends .2 s at 80 mA plus .05 s at 30 mA RX/timeout, with equal success/failure duration and independent failure probability p. r is the allowed additional retries, K maximum attempts, A expected attempts and q synthetic report success probability. Wake and tail happen once; maximum serial duration must fit T.

For p=.2 and r=2, A=1.24 and q=.992. The separate retry ledger gives 36.68825 mC, 132.0777 mJ and 61.1470833333 µA. This does not modify the default registered fixture. At p=1 charge remains finite and q=0; delivered-report energy is unavailable.

Common misconceptionPSM current predicts battery life.

Period charge depends on the whole state distribution. Network wait is not automatically RX-on; distinguish active receive from a timer running during sleep. Peak current and pulse duration need battery-impedance and brownout checks independently of average charge.

Go deeperWhy bytes divided by a headline rate fails here

Scheduling, resource allocation, framing, repetitions, protocol exchanges and firmware policy determine the event. Only the declared 32-byte S32-v1 fixture has timing evidence in this exercise. Custom bytes or unreviewed firmware require a new trace; the workbench refuses linear scaling.

Interactive investigation · Class 1 · illustrative

Cellular Band–Mode–State Matrix

Shortlist an exact configuration, then identify the evidence that blocks a product recommendation. Region-A/B, TEACH bands and C1/C2/C3 are fictional; there are no authorized frequencies or genuine UE categories here.

Guided sequence and complete presets

Follow 1 → 6. Predict before changing each preset. Finish with the owner-assigned memo below. A low energy estimate cannot repair a failed tuple.

Exact capability request

Synthetic card

Network context

Whole tuple / simultaneous request

Evidence packRevised evidence changes identity and source records. It is no real hardware approval; pair it with host-v2.

Requested firmware identity

Requested host revision

Evaluation date (YYYY-MM-DD, UTC)

Separate remote-node event hypothesis · S32-v1

This teaching timeline is an independent node hypothesis. It does not estimate Gateway-C power or transfer a measured trace across radio families. Only 32 bytes and fw-v1/host-v1 have a declared event fixture.

State policy

D0 payload (bytes)32 bytes has a fixture. Any other size returns “timing evidence needed”.

Reporting period (s)

Constant radio supply (V)

Search / timeout duration (s)Used in search, mixed and outage policies. Search success is stipulated only in the successful-search branch.

Total search current (mA)

Search event fraction s (0–1 or unknown)Mixed policy only. Enter a decimal 0–1, or the word unknown. Default .1 is illustrative, not measured coverage.

Additional connected retries r (count)Retry policy only; K = r + 1 complete attempts. Wake and tail occur once.

Independent attempt failure p (ratio)

Separate RF comparison · does not change DC energy

Radiation efficiency η (ratio)Compare with .50 at fixed directivity, match, feed power and propagation.

Committed: 1 · Node-A / T1 · Node-A / Region-A / T1 · fw-v1 / host-v1 · evaluated 2026-09-08 UTC.

Technical result: supported in fixture

All required exact predicates are true and current in this fixture. This is a technical shortlist only.

Requested identities and host RF burden
Tuple	Full configuration
T1	LTE-M-like node; TEACH-EUTRA-M; Synthetic v1; no real TS; TEACH-L; FDD; C1; synthetic resource; PC-L; no genuine UE class; required L-path-v1. Single mode.

Inspect every predicate, source and freshness record

Complete predicate ledger for the committed request
Predicate / tuple	Truth and reason	Evidence context
module / T1	true · module:T1:matched · Exact whole-tuple record: true.	Node-A-module-fixture-v1@1; fw-v1/host-v1; complete; reviewed 2026-09-08, due 2026-12-07
host / T1	true · host:T1:matched · Exact whole-tuple record: true.	Node-A-host-fixture-v1@1; fw-v1/host-v1; complete; reviewed 2026-09-08, due 2026-12-07
network / T1	true · network:T1:matched · Exact whole-tuple record: true.	Region-A-network-fixture-v1@1; fw-v1/host-v1; complete; reviewed 2026-09-08, due 2026-12-07
peer / T1	true · peer:T1:matched · Exact whole-tuple record: true.	peer-v1-peer-fixture-v1@1; fw-v1/host-v1; complete; reviewed 2026-09-08, due 2026-12-07
combination / single mode	true · combination:not-applicable · Single-mode request; no simultaneous support inferred.	Rules v1

Separate release gates: unknown. Carrier owner: real network, provisioning and operator acceptance. Lab owner: applicable qualification and host test deltas. Compliance owner: final market applicability. Antenna owner: mounted RF path, efficiency, isolation and radiated results.

State energy remains a separate question

Committed S32-v1: 32 D0 bytes, 600 s, 3.6 V; registered policy. PSM-only current is 5 µA. The complete event determines period average.

Energy: calculated. Synthetic successful radio event; actual D0 delivery and latency require field evidence. No battery-life prediction.

30.9888 mC · 111.55968 mJ · 51.648 µA average

Peak state current: 80 mA. Inspect pulse duration in the table and battery impedance separately. This average cannot establish brownout immunity.

Expected attempts: 1; synthetic delivery fraction q = 1. Energy per synthetic delivered report: 111.55968 mJ. D0 latency: unavailable.

registered · radio supply ledger
State	Duration (s)	Current (mA)	Charge (mC)
Resume / wake	0.04	50	2
Connected payload transfer	0.2	80	16
Connected / idle tail (once)	2	5	10
Residual PSM-like sleep	597.76	0.005	2.9888

Separate RF comparison: η relative to .50 gives -3.01029995664 dB directional realized-gain change at fixed directivity, match, feed power and propagation. The DC result above is unchanged by η.

Compare the first four requests · canonical fixture
Request	Technical result	Decisive evidence
Node-A / Region-A / T1	supported in fixture	All exact predicates true; actual-network, qualification and market gates remain unknown.
Node-A / Region-A / T2	fail	host:T2:absent-complete
Node-B / Region-A / T2	supported in fixture	All exact predicates true; actual-network, qualification and market gates remain unknown.
Node-B / Region-B / T2	inspect	network:T2:unknown

Inspect the maximum bounded workload: 8 cards × 24 tuples

This missing-evidence diagnostic evaluates 192 requests. Every row must remain inspect. It does not change the committed configuration.

cellular-band-mode-state-matrix/2.0 · p07-m05-capability-rules-v1 · p07-m05-cellular-options-v1. Canonical date 2026-09-08; due 2026-12-07. SI calculations; display rounded to 6–11 decimal places. Independent tolerance: 1e−8 mC, mJ, µA and dB; capability reasons match exactly. No storage, live lookup or automatic evidence repair.

Decision: retain registered, search, mixture and abort branches until a synchronized current/state trace establishes which occurs and how often.

08 / 10

Separate capability from deployed service

A module and host agree. What evidence says the network will expose that mode at the installation?

Provisioning, subscription, operator configuration, roaming and site conditions sit outside a module’s capability table. An acceptance test must exercise the actual firmware/SIM/service combination and record its date. A cached deployment map cannot establish the urgent-command deadline.

Think about itRegion-B has unknown T2 support. Is that equivalent to its explicit false T1 record?

Answer

No. Node-B/T2 stays inspect; obtain the missing dated service evidence. Node-A/T1 fails within Region-B’s synthetic record. Neither state is inferred from signal strength or from a low energy result.

Service evidence to request before the next review
Owner	Specific next evidence	Recheck trigger
Carrier / provisioning	Exact operator, country, SIM profile, subscription, APN/service, mode/band and roaming permission; registration trace.	SIM, service, firmware, network or country change.
Systems / firmware	Granted PSM/eDRX behavior; command release at every reporting phase; unique D0 report timing through core and backhaul.	Timer, protocol, firmware or endpoint change.
Product / service owner	Dated service-continuity and lifecycle commitments for the intended deployment interval; fallback ownership.	Contract, lifecycle notice or planned service end.

D0 latency = queue + access/state wait + transfer + core/backhaul + endpoint processingEach term is seconds on a consistent event timeline; overlap must be modeled explicitly rather than summed twice. A missing or never-delivered event has no successful latency value. Evaluate deadline failures over all generated events.

Go deeperA reachability counterexample without a carrier claim

If an illustrative policy makes the device unreachable for the full 600 s reporting interval, a command released just after sleep starts cannot satisfy a 30 s deadline through that path. This conditional example rejects the policy; it makes no claim about the behavior of an actual operator.

Decision: preserve the carrier evidence gap and date it. No current regional coverage, sunset or roaming claim is made by the fictional matrix.

09 / 10

Understand qualification and host obligations

The module has a certification record. Which product obligations remain at the host boundary?

Start by matching the record to the exact component, firmware, antenna and integration. Then ask the applicable program and authorized lab which evidence can be reused and which host deltas require tests. PTCRB’s current public process separates lab work, operator-specific testing and government requirements. Its PPMD 4.0 change history moved IoT-device requirements to the IoT Network Certified guide; older integration shorthand is unsafe. [PTCRB-PROCESS] [PTCRB-PPMD]

Evidence lanes have different owners
Lane	Required question
Chipset / modem implementation	Which hardware and firmware functions were actually implemented? Vendor statement and capability trace.
Module record	Which exact module, software, RF ports and conditions were evaluated? Match the record and allowed integration.
Final host / antenna	Which host changes affect radiated performance, emissions, RF paths and applicable test deltas? Lab and antenna owners.
Industry program	Which current PTCRB / IoT Network Certified or GCF criteria and test versions apply to this device? Program/lab review.
Operator acceptance	Which carrier/service requires additional acceptance or provisioning? Carrier owner.
Market applicability	Which jurisdictional requirements apply to the final product? Compliance owner, within Path 09.

NAPRD03 V6.25 supplies a program-specific test-reference layer. The GCF public overview provides orientation; detailed member criteria were not available here, and its dated feature table is not treated as a current applicability list. These sources do not close this host’s certification scope. [PTCRB-TEST] [GCF]

Think about itCan changing the enclosure and antenna preserve every conclusion from a module record?

Answer

Only matching, applicable evidence can answer that. The new host configuration creates a review trigger for the antenna and lab owners. It does not erase usable component evidence, but it cannot inherit unproven radiated or market conclusions.

Common misconceptionA certified module makes the host certified.

Record reuse is conditional on the actual program and integration. It does not substitute for final host assessment, carrier acceptance or national applicability. Keep each unresolved lane visible even when all synthetic technical predicates are true.

Go deeperPrepare a reviewable package before asking for a verdict

Collect exact part/firmware identities, antenna specifications, RF-path drawings, enclosure and installation variants, existing records, intended markets and service modes. The lab can then identify concrete deltas. An undifferentiated “global LTE certificate” cannot do that work.

Decision: request a scoped host/qualification plan, with a separate compliance owner. Now assemble the candidate memo.

10 / 10

Write the cellular option decision memo

What can the engineering team decide today, and which observation could reverse that choice?

Conditional shortlist: Node-A/T1 in Region-A is technically supported in the fixture. Node-A/T2 fails its host set; Node-B/T2 in Region-B needs network evidence. Gateway-C/T3 in Region-B is supported singly, while its T1+T3 simultaneous request fails. Real product selection remains open until the dated service, host and qualification evidence matches.

Complete static technology evidence card · available without prior lessons or storage
Field	Snapshot
ID / variant / owner	p07-technology-evidence-card-v1 / P07-M05-CELLULAR-OPTIONS-v1 / RF systems owner
Inherited IDs	P07-M04-REMOTE-FLEET-v1 · P07-M01-LE-EXT-v1 · P07-M02-LE-CONN-v1 · P07-M02-THREAD-SED-v1 · P07-M03-HE-BACKHAUL-v1 · p06-evidence-map-v1 · M01-A · M01-B-LOSS · M01-INSTALL-UNKNOWN
Scenario	1000 fixed remote nodes, metal-machine/open-pole sites and upright/rotated enclosure strata; cellular node or a separately powered cellular gateway. Generic 2.450 GHz QPSK remains a baseline, not cellular.
Requirement	Chosen field goal: ≥99% complete unique 32-byte D0 reports at server D0 within 60 s in each site/orientation stratum, one report/600 s. Urgent commands accompany 1% of reports; ≥99% at node D0 within 30 s of server release at arbitrary reporting phases. No measured achievement.
Family	Separate LTE-M-like node, NB-IoT-like node and NR-like gateway variants; no generic LTE rate or transferable state fixture.
Identity	Synthetic T1/T2/T3 v1 for calculation. Real trace: LTE-M Cat-M1 candidate, TS 36.101 V13.25.0; nRF9160 revision 2 (B1), modem firmware 1.3.7; deployment compatibility unresolved.
Region / device class	Region-A/B are fictional. CH / EU-EEA candidate markets; actual operator, SIM/service, class, band and final configuration need confirmation.
D0 payload / reporting / deadline	256 bits; 600 s period; 60 s uplink deadline.
Latency definition	Complete node D0 report generation to unique complete server D0 receipt; lost/late reports fail, duplicates count once. Commands: server release to node D0, including sleep phase, core/backhaul and processing.
Resources	Whole mode/band/duplex/resource/power-class/RF-path tuple plus a separate simultaneous-combination identity. No scheduled allocation is established for a real network.
States / assumptions	wake: 0.04 s, 0.05 A; connected transfer: 0.2 s, 0.08 A; tail: 2 s, 0.005 A; PSM-like residual: 597.76 s, 0.000005 A; successful search (separate variant): 30 s, 0.04 A. Search is a separate variant that displaces sleep; never append it to an unchanged 600 s base ledger.
RF interfaces	R1 component RF port → feed/match → R2 antenna feed → S0 over-the-air → receive RF chain → R3 decision boundary → D0. D1 is framed data. SI internally; e^(+jωt); count each feed/mismatch/radiation loss once. DC is a separate supply boundary.
Infrastructure	Real operator coverage, provisioning, registration, granted PSM/eDRX timers, roaming, command reachability, backhaul and lifecycle commitments unknown.
Qualification	PTCRB / IoT Network Certified and GCF program scope, exact module record and host test deltas remain open; operator acceptance separate.
Regulatory question	Compliance owner determines market applicability and final host/antenna obligations; Path 09 owns this decision.
Claims / sources / status	CELL-TUPLES: p07-m05-cellular-options-v1 — Synthetic support is conditional; no real-network approval. CELL-CHARGE: p07-m05-cellular-options-v1 — Checked synthetic state ledger, not measurement. CELL-REAL-TRACE: LTE-RF, LTE-CAP, NORDIC-FW, LTE-TEST — Selected clauses read; exact test/version and installed-host evidence incomplete.
Uncertainty	No measured event distribution, current/registration trace, OTA result, country/operator feature commitment, host approval or service lifecycle evidence.
Decision	Shortlist Node-A/T1 in Region-A only within the fixture; Node-B/T2 in Region-B remains inspect. Gateway-C/T3 in Region-B is supported singly; T1+T3 simultaneous fails. Real node/gateway selection remains conditional.
Next evidence	Synchronized supply-current and registration/state trace across successful resume, 30 s search, outage, commands at arbitrary phases and battery impedance extremes, with sequence IDs at D0. A high search fraction or unreachable urgent command can reverse the choice.
Review date / trigger	2026-12-07; Firmware, host/antenna, mode, source revision, operator/SIM, market or service-lifecycle change, or review due date.

Owner-assigned cellular option memo
Owner	Action before a product recommendation
Standards	Close the exact TS/part/version, change-request impact and conformance destination. Record an explicit compatibility edge.
Vendor	Pin order code, firmware and configuration; obtain complete capability and combination records. S32 assumptions are not vendor measurements.
Carrier	Confirm actual service, provisioning, roaming, granted states, urgent-command reachability and dated lifecycle commitments.
Antenna	Verify mounted RF path, isolation and gain/loss convention for the node and each gateway combination; retain the −3.01029995664 dB controlled efficiency comparison.
Lab	Identify applicable program versions, module-record reuse conditions, host deltas and conducted/radiated test conditions.
Compliance	Determine CH / EU-EEA product applicability and authorization obligations independently of the technical shortlist.

Energy basis: keep registered 111.55968 mJ, successful search 4431.01968 mJ, illustrative .1 mixture 543.50568 mJ and abort 4337.45928 mJ per generated report at 600 s and 3.6 V. The abort delivers nothing; the mixture is not a percentile. Gateway power requires its own fixture.

Two rejected assumptions: individually listed bands/resources may be freely combined; PSM current plus a module certificate establishes the battery-powered host’s service readiness. Neither follows from the evidence.

One discriminating next trace: synchronize radio-supply current, registration/state logs and unique D0 sequence IDs during resume, successful search, outage and commands released at arbitrary sleep phases. Include mounted orientations and battery-impedance extremes. A high search fraction or missed urgent commands could reverse the node shortlist and favor a powered gateway architecture.

Go deeperWrite the boundary of the decision in one sentence

“Proceed to a scoped field and host-validation experiment for the exact candidate tuple; do not issue a real go-live approval from these synthetic results.” List the observation that would change this decision and its owner.

Next, 07.6 adds timing, sky view and satellite-fallback questions. A fallback needs its own service and energy evidence; it is not guaranteed by the word satellite.

Ungraded review

Check your understanding

Answer each question in your own words, then reveal the model answer.

01Does “Release 13” identify the RF requirement and its test?
Model answer
No. Pin TS number, part where applicable, version, exact mode, clause and test applicability. The selected TS 36.521-1 V13.4.0 even redirects its requirements to Release 14. A release label alone cannot close that dependency.
02Node-A lists T1 and T2 in its module. Why does T2 fail in Region-A?
Model answer
Its integrated host complete set contains only T1. The exact T2 tuple is absent: host:T2:absent-complete. Independently combining a band, resource and power-class list would invent support.
03Gateway-C supports T1 and T3 singly. Can it use both at once?
Model answer
The separate T1+T3-simultaneous-v1 predicate is explicitly false. Even Both-singles network evidence cannot repair the unsupported combination. Antenna paths, isolation and capabilities need their own combination evidence.
04What changes when one in ten illustrative events searches for 30 s?
Model answer
Search displaces sleep: Q rises from 30.9888 to 1230.8388 mC in that branch. The independent s=.1 mixture is .9×30.9888 + .1×1230.8388 = 150.9738 mC, or 251.623 µA over 600 s. It is a mean, not P95; unknown s yields scenarios only.
05Is Node-B / Region-B / T2 an incompatibility or an evidence gap?
Model answer
It is inspect: network:T2:unknown, with no false technical predicate. Request dated, mode-specific network evidence. Unknown is neither a measured outage nor a pass; an explicit false elsewhere would make the conjunction fail while retaining this unknown.
06What remains after finding a module certification record?
Model answer
Match the exact module/SKU, firmware, antennas and integration conditions; have the lab determine host testing and applicable program versions. Obtain operator acceptance separately. The compliance owner determines final market applicability. A module record alone certifies none of those conclusions.

References and further study

Selected primary material accessed 2026-09-08; review due 2026-12-07 or a document, firmware, host, operator or market change. Legacy revisions are pinned intentionally. Source facts, derived rules and chosen synthetic numbers have separate provenance. Portal access was unreliable; accessible official ETSI editions were used. No current carrier database is embedded.

WORKING · 3GPP / ETSI
TR 21.900 · TSG working methods · 19.1.0 / Release 19. Published: 2026-01. Consulted: §4.0, 4.0A/B, 4.6.1–2, 6.4.1–3, 7.1–3. Informative working-methods report. Document identity, change control and work organization. No product feature or deployment attestation.
LTE-RF · 3GPP / ETSI
TS 36.101 · UE radio transmission and reception · 13.25.0 / Release 13. Published: 2023-07. Consulted: §4.1; Table 5.5-1 Band 20; §5.5E/F, 6.2.2E, 6.2.3E; change history CR 5093. Normative RF requirements; informative history. Selected legacy LTE-M RF navigation example. No complete test plan, numerical receiver limit or host approval inferred.
NR-RF · 3GPP / ETSI
TS 38.101-1 · UE RF, Part 1: Range 1 Standalone · 18.10.0 / Release 18. Published: 2025-08. Consulted: Title, scope and contents. Normative document; scope orientation only here. Identifies the FR1 standalone part for a candidate NR gateway. No NR band, power, CA or sensitivity table used. ETSI directory lists newer 18.14.0; no automatic version migration.
LTE-CAP · 3GPP / ETSI
TS 36.306 · UE radio access capabilities · 13.5.0 / Release 13. Published: 2017-04. Consulted: §4.3.5.1, 4.3.5.1A, 4.3.5.2. Normative capability definitions. Per-band duplex and separate combination capabilities. A field definition is not a capability dump from the candidate product.
LTE-TEST · 3GPP / ETSI
TS 36.521-1 · UE RF conformance testing · 13.4.0 / Release 13. Published: 2017-04. Consulted: §1, 2 and 4. Normative test-document routing. §4 redirects requirements to the Release 14 document. Destination test text and its applicable version were not verified. No test tolerance or conformance pass claimed.
LTE-PHY · 3GPP / ETSI
TS 36.211 · Physical channels and modulation · 13.13.0 / Release 13. Published: 2020-01. Consulted: §5.3.3, 5.6; §10.1.5 heading. Normative waveform definition. PUSCH transform precoding and LTE uplink SC-FDMA scope. No fixed PAPR, PA efficiency or cellular airtime inferred.
NR-PHY · 3GPP / ETSI
TS 38.211 · NR physical channels and modulation · 18.6.0 / Release 18. Published: 2025-04. Consulted: §4.4.5; §6.3.1.4–5; §6.4.1.1.3. Normative waveform / resource definition. BWP and configured transform-precoding branches. Exact scheduling and gateway capabilities unresolved; not asserted to be the newest revision.
LTE-STATE · 3GPP / ETSI
TS 24.301 · EPS NAS protocol · 13.10.0 / Release 13. Published: 2017-07. Consulted: §5.3.11–12. Normative state procedures. Requested PSM/eDRX and network-dependent operation. No current or duration values taken from this standard.
NORDIC-FW · Nordic Semiconductor
nRF9160 modem firmware release notes · mfw_nrf9160 1.3.7. Published: Publication date not established in selected text. Consulted: 1.3.7 intro, supported bands, production and known limitations. Informative vendor implementation statement. LTE-M / NB-IoT firmware; LTE-M Band 20 listed. CH candidate use only, operator/service unverified. Not a standard requirement, supply trace, installed-host result or operator commitment.
NORDIC-SKU · Nordic Semiconductor
nRF9160 downloads · Web page / firmware 1.3.7 target. Published: Undated page, accessed 2026-09-08. Consulted: Modem firmware target revision and certification caveats. Informative vendor compatibility context. Revision 2 (B1) target; revision 1 testing caveat. Exact order code, AT configuration and operator-certified firmware remain a vendor review gate.
PTCRB-PROCESS · PTCRB
Get Certified · Current public process page at access. Published: Undated page. Consulted: Certification process and operator-specific testing questions. Industry-program process. Authorized lab and separate operator/government questions. No final host classification or national authorization.
PTCRB-PPMD · PTCRB
Program Management Document · 4.1. Published: 2025-12. Consulted: §1.2 definitions; change history 4.0 / 4.1. Program requirements. Chipset/module/product roles; IoT requirements routing changed in 4.0. Applicable IoT Network Certified guide and host test scope still need lab confirmation.
PTCRB-TEST · PTCRB
NAPRD03 · Version Specific Technical Overview · 6.25. Published: 2026-06. Consulted: Title, §2.3.3 E-UTRA references. Program test references. Program version selection is a separate dependency. No current candidate certificate or applicable test-list closure established.
GCF · Global Certification Forum
Services / 3GPP wireless devices · Public overview. Published: Undated overview; feature table explicitly January 2024. Consulted: Services introduction and member-information boundary. Informative certification overview. Device certification lane, separate from deployment evidence. Member criteria unavailable; the historical feature table is not a current applicability list.

Model cellular-band-mode-state-matrix/2.0; rules p07-m05-capability-rules-v1; fixture p07-m05-cellular-options-v1. Charge is derived from disclosed total supply-current states. Whole-tuple rules and all numeric anchors have an independent Decimal/set oracle; no source supplies the invented currents, regions or support flags.