Why Your NB-IoT Device Said 10 Years on the Datasheet and Died in 18 Months
June 4, 2026 · 7 min read · Technical Whitepapers
A Huawei Cloud analysis found IoT devices dropping from a theoretical 2 years to 6 months in the field. The gap is not the battery — it is network registration energy that the datasheet math ignores. One re-attach burns as much power as 12 hours of PSM sleep. This article covers the corrected battery math, the AT commands that actually enable PSM, and why a lab measurement always lies.
The marketing math: 5 Wh battery, 3 µA PSM sleep current, 1 transmission per day at 200 mJ. Result: 10.2 years. The field math: same battery, same device, but the network drops the PSM timer to 10 seconds because the visited carrier does not support the home network's T3412 value. The device re-attaches from scratch 4 times per day instead of resuming a session. Each re-attach burns 100-200 mA·s — equivalent to 12 hours of PSM sleep. The device dies in 18 months. The gap between marketing and field is not the battery chemistry, the chipset, or the antenna. It is the PSM timer configuration on a carrier network that was never designed for a device that sleeps 23 hours a day.
The Corrected Battery Math
The standard formula: Battery life = Capacity / (Daily transmit energy + Daily idle energy + Daily sleep energy). The standard mistake: assuming sleep energy dominates. It does not. On NB-IoT, a single network registration event consumes 100-200 mA·s. At 3 µA PSM sleep, that is 9-18 hours of sleep budget burned in 2 seconds of re-attach. If the device re-attaches 4 times per day — because the network drops the session between transmissions — the daily energy budget is dominated by registration, not transmission or sleep.
A corrected calculation for a device transmitting once per hour, 200 bytes per transmission, on NB-IoT Band 8: single transmission energy ≈ 200 mJ, single re-attach energy ≈ 100-200 mA·s (150-300 mJ at 3.6V), PSM sleep between transmissions ≈ 3 µA × 3.6V × 3598 seconds ≈ 39 mJ. If the session persists (no re-attach): daily energy ≈ 24 × (200 + 39) ≈ 5,736 mJ. With 4 re-attaches adding 4 × 200 mJ: daily energy ≈ 6,536 mJ — a 14% increase. If the session drops every transmission (24 re-attaches): daily energy ≈ 24 × (200 + 200 + 39) ≈ 10,536 mJ — an 84% increase. Battery life drops proportionally.
Source: Huawei Cloud, "NB-IoT Device Battery Life Estimation and Power Optimization", November 2025. Available at https://bbs.huaweicloud.com/blogs/469037
The AT Commands That Actually Enable PSM
PSM is not automatically active just because the module supports it. Three AT commands must be sent, and the network must accept the requested values:
| Command | Example | What it does |
|---|---|---|
| --------- | --------- | ------------- |
| AT+CPSMS | AT+CPSMS=1,,"00100001","","10101010","00100001" | Enables PSM; T3324=2s active, T3412=170min extended |
| AT+CEDRXS | AT+CEDRXS=1,5,"0011" | Enables eDRX with ~20.48s cycle (LTE-M) |
| AT*CELLSLEEP | Module-specific | Forces module into lowest-power sleep state |
The network can override requested PSM values. The device requests T3412=170 minutes; the network may grant 10 minutes. The device has no way to know — it enters PSM, the timer expires, it wakes, the session is gone, it re-attaches. The only way to verify actual PSM behavior is to measure current draw at the battery with a tool like Otii Arc (100 nA resolution) over a full transmission+sleep cycle. Do not trust the AT command acknowledgment.
Source: Nexcon.io, "Energy Optimization With PSM and eDRX in IoT Networks", March 2025. Available at https://nexcon.io/blog/energy-optimization-with-psm-and-edrx-in-iot-networks
Non-Steered Roaming: The Battery Saver Nobody Configures
A device on steered roaming always tries the preferred carrier first — even if that carrier's signal at the device location is -115 dBm. The modem transmits at maximum power. Packets are lost. Retransmissions multiply. Each retransmission burns the same energy as the original. A device on non-steered roaming connects to the strongest available network — typically at -85 to -95 dBm. Transmission power drops from 23 dBm to 0-10 dBm. Retransmissions are rare. The connection is stable. PSM timers are maintained because the session does not drop.
The energy difference is not marginal: a device on steered roaming at -115 dBm may consume 3-10x more energy per transmission than the same device on non-steered roaming at -85 dBm — due to the combination of higher TX power, more retransmissions, and more re-attaches. The SIM's roaming policy is a battery life decision, not just a coverage decision.
Source: IoT For All, "Network Selection Behavior Matters: How Non-Steered Roaming Can Improve Battery Life in IoT Devices", 2025. Available at https://dev.iotforall.com/non-steered-roaming-network-selection-behaviors
The Lab vs Field Gap: What to Measure Before Trusting the Datasheet
Three measurements that take 30 minutes and prevent 18-month surprises:
1. Actual PSM sleep current: power the device through a current measurement tool (Otii Arc, Joulescope). Send AT+CPSMS to enable PSM. Wait for the device to enter sleep. Measure current. If it is above 10 µA, PSM is not active — the module is in idle, not sleep. Idle current on some NB-IoT modules is 0.1-2 mA — 100-600x higher than PSM spec.
2. Re-attach frequency over 24 hours: deploy the device in the target location with the target SIM. Log every network attach event. If the device re-attaches more than twice per day, the PSM timer is being overridden by the network or the session is being dropped. Fix the timer negotiation or change carriers.
3. Single-transmission energy: measure total energy for wake → attach → transmit → receive ACK → sleep. Compare against the datasheet value for "single transmission energy." If the measured value is >150% of the datasheet value, signal quality at the deployment location is degrading the link budget. Relocate the device, improve the antenna, or switch carriers.
Source: ITR Blog, "Power-Saving Strategies for Cellular IoT Devices", 2025. Available at https://www.itrvn.com/blogs/power-saving-strategies-for-cellular-iot-devices-part-1