The battery life of the endpoint device is primarily determined by the energy efficiency of the radio protocols of the system, reviewed in a previous article.
Sometimes you can find common mistakes, as in an advertising article, where the main argument in favour of battery life is the following statement: "The devices transmit a radio signal with a power of up to 25 mW. This is 8 times lower than the power emitted by the NB-IoT radio modem." It is incorrect to compare the emission powers to determine autonomy, you have to compare the energy consumed from the battery for the transmission of one message.
Besides the energy efficiency of the radio protocol, there are several other factors that determine the battery life of the ED:
1. Consumption for the transmission of useful information;
2. Consumption of the measuring element of the ED;
3. Consumption for control and service messages;
4. Consumption for down channel operation;
5. Consumption of the ED PCBin stand-by mode.
The consumption of the measuring element of an ED imposes restrictions on the possibilities of using various sensitive and measuring elements in the final device. Besides the fact that the measuring element has to be cheap, it must also consume very little power. Most modern "MEMS" of temperature, humidity, pressure sensors, accelerometers and gyroscopes fall into the right category, you just need to make the right choice and develop appropriate micro-consumption interaction algorithms.
Usually, no more than half of the battery power can be reserved for device operation, so its average consumption should be in the range of 10-20 µA. There is also a set of "huge consumption" devices where the average consumption cannot be reduced for fundamental physical reasons, for example, CO2 or CH4 , or a GPS tracker.
Consumption for service and control messages is especially important for devices close to security topics, when you need to be constantly sure that the sensor is working and you have to periodically send control confirmation messages to the air. These control messages should not consume a significant portion of the battery life. An optimal 'channel check' time in LPWAN would be several times a day.
Down channel is one of the potentially most consuming elements of LPWAN systems, we plan to cover down channel issues in detail in a separate article.
It is also important for device developers to be able to minimise the background consumption of the ED. Many modern microcontrollers provide such a mode, but for its full implementation it is necessary to properly control the processor.
NB-IoT is based on synchronous communication with high overhead and quota requirements, which is very draining of the battery power. The energy efficiency of ISM band systems as LoRaWAN and SigFox is many times better than that of NB-IoT.It is known from publicly available sources that the SigFox protocol allows you to send 15-20 thousand messages from the single 3.6 V, 2 200 mAh battery. The standard GoodWAN device - transmits around 50,000 messages from the same single 2,200mAh battery, providing 5 years of battery life for the device when transmitting a message once per hour. Read more about the energy efficiency of LPWAN systems in our Habr article.
- In all radio networks there is always the probability of an individual message being lost.
- The asynchronous nature of message transmission always implies a non-zero probability of collisions over the air.
- Delivery of a message with receipt in LPWAN networks is hardly realised due to the down-channel characteristics.
The GoodWAN network is built based on the probability of loss of one non-critical message 10-1 and provides a probability of loss for critical messages 10-5.
Comparing the out-door coverage area in a real city environment for different types of LPWAN systems is most easily evaluated in terms of the probability of missing a message depending on the distance.
The number of messages transmitted and received depending on the distance to the gateway in real urban areas for outdoor coverage.
The following graph shows the results of comparative tests of GoodWAN and LoRaWAN(Actility) gateways in a real city environment in Moscow. The gateways were installed in the same conditions on a 15-storey building and their antennas were periodically swapped. The end units operated at a transmission frequency of 10 seconds and were installed in a car which drove along the city streets along random routes within a radius of 3 to 9 kilometres from the gateways. The total number of messages transmitted (theoretical reception limit) and the number of messages actually received by each gateway - depending on the distance - were monitored.