Sensor selection conversations in IoT deployments tend to focus on the physical layer: measurement accuracy, operating temperature range, ingress protection rating, and power draw in active and sleep states. These are legitimate engineering concerns. But the connectivity requirements that a given sensor type imposes on the broader system architecture are just as consequential, and they are often addressed too late in the design process.
The wrong connectivity choice for a sensor category doesn’t always manifest as a hard failure. More often it shows up as degraded system performance. A temperature sensor that should transmit every fifteen minutes instead buffers for hours because NB-IoT session establishment overhead makes frequent transmissions cost-prohibitive. A vibration sensor generating high-frequency waveform data saturates a low-bandwidth LPWAN link. A GPS asset tracker burns through battery reserves because the chosen radio technology requires too much uplink time to acquire a network. These are architecture mismatches, and most of them are preventable if connectivity requirements are mapped to sensor type before hardware selection locks in.
This piece works through the main IoT sensor categories, the data characteristics each one generates, and the connectivity technology trade-offs those characteristics drive. The intent is to give architects and deployment engineers a practical framework for aligning sensor selection with connectivity infrastructure from the start.
How Sensor Data Characteristics Drive Connectivity
Before going sensor by sensor, it is worth establishing the four data characteristics that determine connectivity requirements for any IoT sensor. Each one influences technology selection independently, and the combination of all four defines the full connectivity requirement profile for a given deployment.
Payload size is the raw volume of data generated per transmission event.
- Binary state sensors (door open/closed, tamper detect) generate payloads measured in single-digit bytes
- Environmental sensors transmitting a single reading generate 10 to 50 bytes per event
- Vibration sensors transmitting FFT spectral data generate payloads ranging from several kilobytes to hundreds of kilobytes per transmission
- Video and imaging sensors generating continuous streams operate at different orders of magnitude
Payload size drives minimum bandwidth requirements and directly affects the viability of LPWAN technologies, which impose strict maximum message size limits at the protocol level.
Transmission frequency is how often the sensor needs to deliver data to the platform.
- Utility metering transmitting four readings per day sits at the extreme low end
- Environmental monitoring at one-minute intervals represents a moderate frequency use case
- Machine condition monitoring transmitting at one-second intervals or faster is at the high end of what most LPWAN technologies can handle without performance degradation
Transmission frequency, combined with payload size, determines total throughput requirements. It also determines whether session establishment overhead becomes a meaningful cost in latency and power budgets.
Latency tolerance is how long the application can wait between data generation at the sensor and data receipt at the platform.
- Soil moisture sensors in precision agriculture can tolerate latency measured in minutes or hours with no operational consequence
- Cold chain temperature monitoring typically tolerates latency in the range of one to fifteen minutes
- Patient-worn cardiac monitors and safety-critical industrial sensors require near-real-time delivery with latency measured in seconds
Latency tolerance determines whether store-and-forward architectures are viable and gates the use of technologies like NB-IoT that carry higher session establishment latency than LTE-M or 4G.
Mobility profile describes how the sensor moves during its operational life.
- Fixed sensors installed permanently at a known location have no handoff requirements and no multi-carrier coverage needs
- Occasionally mobile sensors such as assets that move periodically between depots need connectivity that survives infrequent location changes
- Continuously mobile sensors on vehicles, personnel, or shipping containers in transit need session continuity across cell tower handoffs, carrier coverage zones, and potentially international borders
Mobility profile drives decisions about network technology selection, SIM architecture, and whether multi-network or eUICC-based connectivity is required.
With these characteristics as the evaluation framework, sensor-specific connectivity requirements become considerably more tractable.
Environmental and Condition Monitoring Sensors
Temperature, humidity, pressure, air quality, and gas detection sensors are among the most widely deployed sensor types across industrial IoT, cold chain logistics, smart buildings, and utilities applications. Their data profile is well-matched to LPWAN technologies.
Typical data characteristics for this category:
- Payload size: 10 to 50 bytes per reading
- Transmission frequency: once per minute to once per hour depending on application
- Latency tolerance: high, typically minutes to hours
- Mobility profile: fixed or near-fixed
NB-IoT is the technology recommendation for this category. Its power efficiency in eDRX and PSM configurations allows battery-operated environmental sensors to achieve multi-year field life without external power. Its superior link budget relative to LTE-M makes it viable for sensors in locations where above-ground cellular performance is attenuated, including basement mechanical rooms, underground utility vaults, and dense industrial enclosures.
LTE-M is the preferred alternative when any of the following conditions apply:
- The deployment requires OTA firmware updates on a regular cadence, where NB-IoT’s narrow bandwidth allocation makes large downlink payloads impractical
- Transmission frequency is high enough that NB-IoT’s session establishment latency creates unacceptable power overhead per transmission cycle
- The application requires voice or full-duplex communication alongside sensor data
The key architecture consideration for environmental sensors on NB-IoT is session establishment cost. Devices in PSM spend the majority of their operating life in a state where the network considers them unreachable. Each transmission cycle requires waking, re-establishing a network session, transmitting the payload, and returning to PSM. At sub-minute transmission frequencies, this overhead can make NB-IoT energetically and economically unviable, and LTE-M or 4G Cat-1 becomes the more appropriate choice.
Vibration, Acoustic, and Machine Condition Monitoring Sensors
Industrial condition monitoring is among the most connectivity-demanding sensor categories in IoT. Vibration sensors used for bearing fault detection, acoustic sensors monitoring for anomalous machine signatures, and current sensors tracking motor behavior generate data at sampling rates that are architecturally incompatible with the small-payload, low-frequency model that LPWAN technologies are designed for.
Typical data characteristics for this category:
- Payload size: kilobytes to hundreds of kilobytes per transmission for spectral data; megabytes for raw waveform capture
- Transmission frequency: seconds to minutes during active monitoring periods
- Latency tolerance: moderate to low depending on whether the application is predictive maintenance or real-time anomaly detection
- Mobility profile: fixed
Connectivity technology options and their trade-offs:
- NB-IoT is viable only for transmitting pre-processed summary statistics, not raw or spectral data. The technology’s bandwidth ceiling and message size constraints make it unsuitable for full waveform or high-resolution FFT payloads.
- LTE-M is viable when edge processing moves FFT analysis to the sensor or gateway level, reducing payloads to derived spectral features rather than raw data. The connectivity cost savings of LTE-M versus 4G Cat-4 across thousands of sensors can justify significant investment in edge compute hardware.
- 4G Cat-4 is the baseline recommendation for full waveform or high-resolution spectral data transmission without edge processing. The higher throughput handles payload volume without constraining transmission frequency.
- 5G NR is the appropriate choice for deployments requiring real-time waveform analysis, sub-second latency, or high sensor density in a shared spectrum environment.
- Private LTE or private 5G is increasingly the right architecture for dense IIoT condition monitoring deployments where data volume is high, shared spectrum contention is a concern, and the facility has an operational technology security requirement to keep industrial data on-premises.
The edge compute decision is the most consequential architecture trade-off in this category. Running FFT analysis at the edge shrinks payloads dramatically and enables the use of lower-cost, lower-power connectivity technologies. The up-front investment in edge compute hardware often pays back quickly at scale when connectivity cost per device is reduced across a large sensor population.
GPS and Location Tracking Sensors
GPS and GNSS-based location sensors span a wider range of connectivity requirements than almost any other sensor category, because deployment scenarios vary from low-power pallet trackers transmitting twice a day to high-frequency vehicle telematics devices transmitting concurrently across multiple data streams.
Low-power asset tracking covers pallets, containers, high-value equipment, and returnable assets.
- Payload size: small, typically latitude, longitude, timestamp, and battery status (under 100 bytes)
- Transmission frequency: minutes to hours between updates
- Latency tolerance: high
- Mobility profile: occasionally mobile, with infrequent location changes
LTE-M is the recommended technology for this use case. PSM and eDRX support enables multi-month or multi-year battery life on compact form factors, while higher throughput versus NB-IoT makes periodic OTA configuration updates practical. NB-IoT is viable for the most power-constrained, lowest-frequency cases, particularly for near-fixed assets in strong NB-IoT coverage environments, but its session establishment latency creates problems for applications requiring rapid location acquisition on motion trigger.
High-frequency vehicle telematics covers commercial fleet tracking, emergency vehicle monitoring, and heavy equipment management.
- Payload size: moderate to large, with concurrent GPS, engine diagnostic, ELD, and driver behavior event streams
- Transmission frequency: 30 seconds or faster for position updates; near-continuous for event-driven data
- Latency tolerance: low for safety-critical events; moderate for routine position updates
- Mobility profile: continuously mobile across carrier coverage zones and potentially international borders
LTE-M Cat-M1 is the minimum viable technology for this use case. 4G Cat-1 or Cat-4 is the appropriate choice for deployments with heavier concurrent data streams.
Multi-network SIM architecture is important for mobile location tracking. A telematics device locked to a single carrier will encounter coverage gaps on any route of meaningful geographic diversity. Multi-IMSI SIM solutions, or eUICC-based eSIM with OTA profile switching via RSP, allow the device to connect to the strongest available network without manual re-provisioning. For cross-border fleet operations, this is the difference between continuous telematics reporting across a border crossing and a vehicle that goes dark until a dispatcher manually investigates.
Flow, Level, and Utility Metering Sensors
Smart metering across water, gas, and electricity utilities, along with flow and level monitoring in industrial and infrastructure applications, represents one of the largest deployed IoT sensor segments globally. The data profile of this category is well-suited to LPWAN.
Typical data characteristics:
- Payload size: very small, typically under 50 bytes per reading
- Transmission frequency: four to twenty-four times per day for most metering applications
- Latency tolerance: very high, hours to days in many cases
- Mobility profile: fixed
NB-IoT is the dominant technology recommendation for utility metering at scale, for several compounding reasons:
- Its link budget advantage over LTE-M makes it viable in underground meter pits, basement installations, and dense urban canyon environments where above-ground cellular performance is significantly attenuated
- Per-SIM data costs for small, infrequent payloads are lower on NB-IoT than LTE-M, and at city-wide or regional deployment scale the per-unit cost difference aggregates to meaningful total cost of ownership savings
- NB-IoT’s lower power consumption at very low transmission frequencies extends battery life beyond what LTE-M delivers for the same duty cycle
The architecture consideration specific to utility metering is deployment longevity. Smart meters commissioned today will be in service for ten to fifteen years. The cellular network landscape at end of life will differ from the one at deployment, as 2G and 3G sunset timelines have already demonstrated. eUICC with remote SIM provisioning via SGP.02 or the IoT-optimized SGP.32 specification is not an optional feature for infrastructure with decade-scale field life. It is the mechanism that allows network profile updates to be pushed OTA without physical device access, which is the only viable remediation path for a large fixed sensor estate when a carrier transitions or a network technology is deprecated.
Imaging, Video, and High-Bandwidth Sensors
Camera-based sensors covering surveillance, machine vision for quality control, traffic monitoring, and remote site security represent the high end of the IoT connectivity bandwidth spectrum. The data volume generated by even compressed video at modest resolutions is different from the sensor types discussed above.
Connectivity requirements for this category:
- Payload size: large to very large, from megabytes per event for triggered clips to continuous multi-megabit streams for live video
- Transmission frequency: continuous for streaming; event-driven for clip-based architectures
- Latency tolerance: low for real-time applications; moderate for triggered clip upload
- Mobility profile: generally fixed, with some exceptions for mobile surveillance and vehicle-mounted cameras
Technology selection by architecture type:
- Event-triggered clip upload with edge inference: LTE-M Cat-M1 can be sufficient for the uplink channel when the sensor runs inference locally and transmits only metadata or short triggered clips. Downlink requirements for firmware updates and model deployment still favor LTE-M over NB-IoT.
- Continuous video streaming: 4G Cat-4 is the minimum viable technology. The sustained throughput requirements of even low-resolution continuous streaming exceed what LTE-M can reliably deliver at scale.
- Real-time machine vision with sub-10ms latency requirements: 5G NR with edge compute co-located at the base station is the architecture that makes this use case viable. The combination of 5G’s latency profile, multi-gigabit throughput in mmWave deployments, and edge inference at the network edge enables real-time quality control and autonomous systems applications that are not practical over 4G.
Power architecture is the practical deployment constraint for this category. 5G and 4G Cat-4 radios consume power at levels that make battery operation for continuous streaming non-viable beyond short windows. High-bandwidth sensor deployments are almost universally line-powered, which constrains deployment locations to where power infrastructure exists or can be economically extended.
Matching Connectivity Infrastructure to Sensor Architecture
Sensor selection and connectivity technology selection are inseparable architecture decisions. A deployment that optimizes sensor hardware without mapping connectivity requirements to each sensor category will encounter performance and cost problems that are difficult and expensive to remediate once devices are in the field.
The evaluation framework is consistent across all sensor types:
- Payload size and transmission frequency determine minimum bandwidth requirements and drive the LPWAN versus broadband decision
- Latency tolerance determines whether store-and-forward or PSM architectures are viable
- Mobility profile determines multi-network SIM requirements and handoff handling
- Deployment scale and field life determine whether eUICC with OTA provisioning is required rather than optional
KORE’s connectivity infrastructure supports the full technology stack this sensor-by-sensor analysis points to, including NB-IoT, LTE-M, 4G, 5G, LoRaWAN, and satellite, across 40+ carrier partnerships and 200+ countries. The ConnectivityPro platform provides SIM lifecycle management, network policy controls, and usage governance that enterprise deployments require when the device estate spans multiple sensor categories, connectivity technologies, and geographies simultaneously. With 21M+ active SIMs deployed across industrial, health, fleet, and utilities verticals, the infrastructure is built to support sensor deployments from pilot scale to global production.
The sensors generate the data. The connectivity architecture determines whether that data arrives reliably, cost-effectively, and at the scale the deployment demands.
Ready to align your sensor deployment with the right connectivity architecture? Talk to a KORE connectivity engineer about building an IoT infrastructure that matches your sensor types, data requirements, and deployment scale.
Published June 9, 2026.