Comparing LoRaWAN Spreading Factor Airtime Costs against Cellular Repetition Overhead

Payload

Transmitting twenty bytes of telemetry across an unguided radio link requires translating binary data into modulated RF symbols. The duration of that transmission defines the power budget, network capacity, and co-channel interference profile of the remote terminal. In LoRaWAN networks, link budget extension relies on chirp spread spectrum modulation, where increasing the spreading factor doubles the chirp duration per chip.

In cellular narrowband networks such as NB-IoT (LTE Cat-NB1/NB2) and LTE-M (LTE Cat-M1), link budget extension relies on physical layer retransmissions across subframes. Both approaches achieve range extension at the expense of time on air, yet the mathematical relationship governing that time on air differs fundamentally between chirp modulation and cellular subframe repetition.

Airtime Scaling across Chirp Modulation Rates

LoRa physical layer framing scales transmission duration through log-base-two steps. Each unit increase in spreading factor from SF7 to SF12 halves the effective data rate and doubles the symbol duration. The basic symbol period equals two raised to the power of the spreading factor divided by the modulation bandwidth.

At a standard bandwidth of 125 kHz, a Spreading Factor 7 symbol lasts 1.024 milliseconds. The link margin decays rapidly. Elevating the link to Spreading Factor 12 expands the symbol period to 32.768 milliseconds.

A transmission containing twenty payload bytes consumes approximately 56 milliseconds of total airtime at SF7 under an explicit header setting with coding rate 4/5. That identical payload sent at SF12 expands to 1482.8 milliseconds of continuous radio output. Uplink airtime dictates battery life.

The receiver sensitivity gain achieved by this expansion follows a linear curve in the logarithmic domain. SF7 offers a typical sensitivity threshold of -123 dBm at 125 kHz bandwidth. SF12 lowers the receiver sensitivity floor to -137 dBm.

The physical layer gains 14 dB of additional link budget. Achieving that 14 dB gain requires multiplying the total airtime by a factor of twenty-six. The power amplifier remains active throughout this extended duration, turning physical layer range gain into an energy expenditure challenge.

Repetition Mechanics in Narrowband Cellular Transport

3GPP specifications define physical layer retransmissions to bridge sub-surface path loss. Rather than slowing down the symbol modulation rate, NB-IoT and LTE-M retransmit identical transport blocks over sequential time slots. NB-IoT NPUSCH format 1 permits repetition counts ranging from 1 to 128 allocations under Coverage Enhancement Level 2.

LTE-M utilizes Coverage Enhancement Mode A for moderate path loss and Mode B for deep attenuation, supporting up to 2048 repetitions on the physical uplink shared channel.

A single subframe in LTE-M lasts 1 millisecond, while an NB-IoT resource unit slot duration varies based on subcarrier spacing. For 15 kHz single-tone allocation, an NB-IoT resource unit consumes 8 milliseconds. Transmitting a 20-byte transport block without repetitions requires one resource unit, yielding an on-air duration of 8 milliseconds.

Applying a 16-repetition schedule expands the physical transmission time to 128 milliseconds. Pushing the module to 128 repetitions under extreme path loss conditions drives the continuous RF active transmission time to 1024 milliseconds. The link margin expands proportionally, raising the maximum coupling loss from 144 dB to 164 dB for NB-IoT.

SF12 transmission of a twenty-byte payload requires 1482 milliseconds of airtime at 125 kHz bandwidth, yielding a 14 dB gain in receiver sensitivity over SF7 at a twenty-six-fold energy penalty.

Comparing the raw transmission times reveals distinct scaling behavior. LoRaWAN SF12 achieves an airtime of 1482.8 milliseconds for a -137 dBm sensitivity floor. NB-IoT at 128 repetitions generates 1024 milliseconds of uplink active airtime to maintain a maximum coupling loss of 164 dB.

Although the total time on air appears comparable in the deep coverage regime, the current draw during those milliseconds differs dramatically due to power amplifier transmit levels and cellular baseline circuitry. Selecting an inappropriate spreading factor or repetition factor forces premature battery replacement or causes persistent packet loss in field deployments.

Frame

Data encapsulation dictates the structural efficiency of every wireless transmission. Raw sensor measurements never travel alone across an RF channel. Protocol layers append preamble sequences, MAC headers, routing metadata, security tokens, and forward error correction parity bits.

The ratio of user payload bytes to total transmitted bytes dictates the actual energy overhead per bit of information transferred.

Header Expansion across Layered Protocol Stacks

Unicast telemetry relies on layered encapsulation to guarantee node addressing and data integrity. LoRaWAN utilizes an unsegmented, highly optimized frame format designed specifically for short telemetry packets. The physical layer adds an unmodulated preamble, a sync word, and an explicit physical header with CR 4/5 protection.

The MAC layer appends a 1-byte frame control field, a 2-byte frame counter, a 4-byte device address, and a 4-byte message integrity code (MIC). A 20-byte user payload sent via LoRaWAN Class A results in a total MAC payload of 33 bytes, yielding a protocol overhead of 39 percent relative to total frame size.

Cellular protocol stacks carry deep legacy overhead originating from broadband networking specifications. An NB-IoT packet must pass through the Packet Data Convergence Protocol (PDCP), Radio Link Control (RLC), Medium Access Control (MAC), and Physical (PHY) layers. If IP networking is selected, an IPv6 header adds 40 bytes, while UDP adds an additional 8 bytes.

Transmitting 20 user bytes over UDP/IPv6 creates a 68-byte transport unit before physical layer framing occurs. Using Non-IP Data Delivery (NIDD) eliminates the IP/UDP header, passing raw payloads directly through the Service Capability Exposure Function (SCEF) or Control Plane CIoT EPS Optimization. Under NIDD, cellular RLC/MAC headers and security framing still add approximately 10 to 14 bytes of encapsulation.

Preamble Sequences and Synchronization Costs

Receivers require initial reference patterns to lock symbol timing before decoding payload bytes. LoRaWAN specifications establish a default preamble length of 8 symbols. At SF7, these 8 preamble symbols consume 8.19 milliseconds.

At SF12, those same 8 preamble symbols consume 262.14 milliseconds before the preamble detector passes timing lock to the frame synchronization engine. Header bytes expand total airtime. The physical preamble alone at SF12 consumes more energy than an entire SF7 data frame.

Cellular networks mandate synchronization across both frequency and time domains to maintain orthogonal subcarrier spacing. An NB-IoT terminal must process the Narrowband Primary Synchronization Signal (NPSS) and Narrowband Secondary Synchronization Signal (NSSS) before transmitting on the Narrowband Physical Random Access Channel (NPRACH). If the terminal has detached or dropped from Connected Mode, transmitting a 20-byte frame demands establishing an RRC Connection.

This control plane exchange consumes multiple uplink and downlink messages before user payload transmission begins.

Structural liabilities in framing design present severe operational penalties when deployed in weak signal environments:

• Padding Wastage in cellular Transport Block Size (TBS) allocations forces small payloads to round up to fixed bit allocations, wasting active power on empty frame padding.
• Preamble Symbol Inflation under high LoRaWAN spreading factors multiplies baseline symbol durations, consuming up to eighteen percent of total packet airtime purely on synchronization.
• Security Token Overhead from 3GPP NAS ciphering and LoRaWAN AES-128 MIC tokens consumes non-negotiable byte fields regardless of how small the sensor measurement is.
• Control Plane Signalling Loops during cellular RRC establishment generate up to ten times the energy drain of the underlying application data payload when devices wake from deep sleep.

Control plane NIDD transmission reduces cellular header overhead from fifty-six bytes down to fourteen bytes, doubling protocol payload efficiency for sub-fifty byte telemetry frames.

Header expansion degrades power efficiency rapidly when the sensor payload is small. A 10-byte temperature report sent over UDP/IPv6 on an LTE-M link results in a protocol efficiency below fifteen percent. The transmitter spends eighty-five percent of its operational energy moving networking headers and synchronization structures.

Module vendors frequently explain that transport layer overhead remains negligible compared to base station synchronization routines.

Drain

Battery degradation tracks total milliamperes spent per message transaction. Calculating actual energy draw requires measuring active transmit current, active receive current, idle listening states, and sleep current over time. A simple comparison of time on air yields inaccurate battery projections if power amplifier current draw and receiver window profiles are ignored.

Will Heavy Repetitions Exhaust Lithium Thionyl Chloride Batteries?

Primary battery chemistry faces severe passivation and voltage breakdown under high peak current pulses. Lithium Thionyl Chloride (LiSOCl2) cells are favored in utility metering for their low self-discharge rate and high energy density. However, LiSOCl2 batteries exhibit high internal resistance.

When a cellular module powers its transmitter to +23 dBm under Coverage Enhancement Level 2, current draw spikes to between 180 mA and 320 mA depending on supply voltage and antenna impedance. High spreading factor LoRaWAN devices transmit at +14 dBm in European ISM bands or +20 dBm in North American allocations, drawing between 45 mA and 120 mA.

Cellular modules draw heavy pulses. Subjecting a bobbin-type LiSOCl2 battery to a 300 mA pulse lasting 1024 milliseconds causes an instantaneous voltage drop. If the voltage drops below the power management IC threshold, the module resets mid-transmission.

Mitigating this failure mode requires placing a hybrid layer capacitor (HLC) or supercapacitor in parallel with the primary battery, adding material cost to the power supply stage.

LoRaWAN transmitters operating at +14 dBm draw substantially lower peak currents. A 45 mA current pulse for 1482 milliseconds imposes significantly less transient load on the battery chemistry, avoiding severe voltage drop and removing the strict requirement for parallel supercapacitors in basic telemetry hardware.

Active Listen Windows and Idle Current Profiles

Uplink cycles represent only a fraction of total terminal energy expenditure. Class A LoRaWAN end-devices open two receive windows (RX1 and RX2) after every uplink transmission. RX1 opens precisely one second after uplink completion, followed by RX2 two seconds after uplink completion.

The transceiver enters a receive state drawing approximately 10 mA to 16 mA for a few milliseconds per window. If no downlink preamble is detected, the radio immediately returns to a sleep state drawing less than 2 microamps.

Cellular terminals remain bound to complex network state machines. Following an uplink event, an NB-IoT module enters RRC Connected mode. The base station holds the terminal in this state using an active RRC Inactivity Timer, typically set between 10 and 20 seconds by network operators.

During RRC Connected state, the module receiver remains active, consuming 40 mA to 80 mA while awaiting network release commands. Once released, the device transitions to Extended Discontinuous Reception (eDRX) or Power Saving Mode (PSM).

Holding an NB-IoT module in RRC Connected mode for ten seconds following a transmission consumes up to 500 mAs of charge, dwarf-scaling the energy spent on the actual physical airtime of a single packet.

The energy budget of an NB-IoT device operating under good link conditions is dominated by the RRC Inactivity Timer rather than physical layer airtime. A 1x repetition transmission consumes 0.96 mAs during transmit, but the subsequent 10-second idle hold draws 500 mAs. Peak currents drop battery voltage.

In contrast, a LoRaWAN SF12 transmission spends 66.73 mAs on transmit airtime, with RX window costs adding less than 0.15 mAs. LoRaWAN energy consumption scales directly with spreading factor, whereas cellular energy consumption is heavily dictated by operator network timers. Engineers treat peak current mitigation as the primary sizing parameter when matching primary cell chemistries to repeating cellular transmitters.

Toll

Radio frequency regulations impose hard physical ceiling caps on spectrum occupation. Unlicensed ISM bands share spectrum among millions of uncoordinated transmitters, requiring regulatory authorities to enforce strict airtime ceilings. Licensed cellular spectrum faces no statutory duty-cycle limits, but network operators enforce commercial and bandwidth management tolls to prevent cell congestion.

Regional Spectrum Regulations and Airtime Caps

National telecom authorities enforce maximum transmission times across unlicensed sub-gigahertz allocations. In Europe, ETSI EN 300 220 standards divide the 868 MHz band into distinct sub-channels with mandatory duty-cycle limits. The primary sub-band (868.1 to 868.5 MHz) mandates a 1 percent duty-cycle cap.

A terminal may transmit for a maximum of 36 seconds per hour within this allocation.

Duty cycle limits enforce pauses. Transmitting an SF12 packet lasting 1482.8 milliseconds consumes 4.1 percent of the total hourly airtime allowance in a single uplink. Sending nine SF12 packets within one hour exhausts the legal 36-second transmission ceiling, forcing the device to silence its radio for the remainder of the hour.

High spreading factors restrict message frequency. Conversely, an SF7 packet lasting 56.6 milliseconds permits sending up to 636 messages per hour within the same 1 percent duty-cycle boundary.

In North America, FCC Part 15.247 governs the 915 MHz ISM band. The FCC does not mandate a percentage duty-cycle limit, but specifies a maximum channel dwell time of 400 milliseconds if the system utilizes fewer than 50 hopping channels. LoRaWAN 125 kHz uplink channels in the US band plan use 64 channel allocation schemes, satisfying the 50-channel rule and permitting extended packet durations.

However, hybrid operating modes or 500 kHz downlink channels remain strictly bound to maximum dwell time boundaries.

Cellular Data Tariffs and Transmission Increments

Commercial mobile networks meter data transactions using fixed minimum byte boundaries. Cellular operators bill NB-IoT and LTE-M connections based on monthly megabyte allowances. A terminal transmitting 20 bytes of application data every hour generates 14.4 kilobytes of raw payload per month.

However, carrier billing engines apply minimum data session rounding increments, often setting minimum session sizes to 1 KB or 10 KB.

If a cellular device opens a new Packet Data Network (PDN) connection for every transmission, a 20-byte write rounds up to the 1 KB billing threshold. The monthly billable data volume jumps from 14.4 kilobytes to 720 kilobytes. Maintaining an active socket connection avoids repeated PDN establishment penalties, but demands sending periodic keep-alive messages.

These keep-alives consume active current and increase cellular data airtime.

Executing an operational audit of module link parameters requires verifying airtime compliance prior to hardware commitment:

1. Calculate raw payload byte length including application metadata and security signatures.
2. Determine required receiver sensitivity based on site survey path loss measurements.
3. Select target modulation profile matching target sensitivity threshold.
4. Verify total frame airtime against regional regulatory ceiling limits.
5. Calculate peak current pulse duration against primary cell voltage drop parameters.

ETSI EN 300 220 Clause 7.2.3 enforces a strict one percent duty cycle ceiling that restricts devices on 868.0 to 868.6 MHz to thirty-six seconds of cumulative uplink per hour.

Fleet

Distributed end-node populations encounter varied propagation environments across field installations. Deploying thousands of smart meters or asset trackers across a city guarantees a statistical spread of path loss values. Antenna placement, building penetration, and ground clearance create a multi-tier link margin profile across the fleet.

Gateway and Cell Site Scalability Limits

Base station radios handle concurrent channel access through orthogonal code or frequency division. LoRaWAN gateways employ multi-channel chips capable of demodulating eight or sixteen channels simultaneously across multiple spreading factors. Spreading factors are largely orthogonal to each other.

A gateway demodulates an SF7 packet and an SF12 packet occurring concurrently on the same frequency channel without frame collision.

Packet collisions destroy radio capacity. High spreading factor airtimes disrupt this capacity model. Because an SF12 packet occupies the channel for 1482 milliseconds, it increases the probability of collision with other transmissions using the same spreading factor.

If ten percent of a 10,000-device fleet falls into weak coverage areas requiring SF12, those 1,000 devices can consume over eighty percent of the gateway’s total channel occupancy. The spatial capacity of the network degrades, reducing total message throughput for nearby SF7 terminals.

Cellular networks manage capacity through centralized scheduling at the eNodeB base station. NB-IoT channels use narrowband physical uplink shared channels (NPUSCH) dynamically scheduled in time and frequency. When a terminal enters Coverage Enhancement Level 2 requiring 128 repetitions, the eNodeB must reserve physical subframes for over a full second per transmission.

Network capacity declines under repetitions. If multiple terminals demand maximum repetitions simultaneously, the cell site experiences physical uplink channel congestion, delaying scheduling grants and increasing device wait times.

Long Term Operating Cost Tradeoffs

Total cost calculations integrate hardware acquisition fees with ongoing field servicing and connectivity rates. Evaluating protocol selection across multi-year operational lifetimes requires balancing silicon costs, battery sizing, and carrier subscription charges.

Selecting the optimal wireless access technology involves evaluating key operational trade-offs across deployment scenarios:

• Unlicensed Sub-GHz Networks operate without monthly carrier subscription fees, but demand private gateway deployment and ongoing infrastructure maintenance.
• Public Cellular IoT Networks eliminate local gateway maintenance costs, but introduce recurring SIM monthly charges and carrier acceptance approvals.
• High Spreading Factor Terminals require larger capacity primary batteries, raising initial bill-of-materials costs while limiting maximum transmission frequency.
• Cellular Coverage Enhancement Terminals provide superior building penetration up to 164 dB MCL, but require power supply designs capable of handling high peak current pulses.

The total cost landscape demonstrates that link budget physics directly dictates deployment economics. High spreading factor airtime on LoRaWAN networks imposes an energy cost that requires larger battery capacities and increases gateway density. Cellular retransmission overhead drives up hardware costs due to energy supply demands and introduces ongoing connectivity fees.

Engineers balance edge-case link margins against cumulative channel occupancy to protect battery longevity and maintain total network capacity.