Originally published on: www.linkedin.com
The article was written by Actility in collaboration with www.orange.com
P.S. The intent of this post is not to say which technology is better or worse. But, rather to show how Cellular IoT and LoRaWAN complement each other. At Actility, we deliver solutions that cater to both LoRaWAN and Cellular IoT.
Battery lifetime is by far one of the most important considerations for LPWAN applications. The reason being that when you deploy loT of sensors, it is very expensive to replace batteries physically and will ruin the ROI if the end-devices last only few years. That is the reason every technology whether LoRaWAN or Cellular IoT (Cat-M1, NB-IoT) all seem to claim the battery lifetime of 10 years but do not clearly state the traffic pattern of the device and how far the device is located from the base station/Gateway. In this article, I shed some light on how LoRaWAN has much lower power consumption to NB-IoT. More importantly, NB-IoT can address more premium use cases that involve higher throughput that cannot be addressed by LoRaWAN. There is no right or wrong answer, the requirements of the application are what determine which radio connectivity to use.
Closer Look at the Access Protocol behind LoRaWAN and NB-IoT
Some of the IoT applications such as asset tracking use rechargeable devices and have a battery lifetime anywhere from 7 to 30 days, but there are applications in which device is deployed in hard to reach areas and need battery lifetime of 10+ years (for example smart gas meters located in indoor basements). In general, LoRaWAN class A device uses minimal power consumption due to the simplicity of the radio and the fact that device is only active when transmitting and is sleeping most of the time (as shown in Fig. 1a). This is unlike cellular technologies in which device has to periodically wake up to synchronize to the network even if it has no data to transmit (as shown in Fig. 1b).
As shown in Fig. 1b, each uplink payload in NB-IoT has numerous TX/RX/Idle transitions due to the complexity of the protocol. The asynchronous nature of LoRaWAN makes the modem design very simple and cheap to implement, however cellular technologies employ advanced scheduling algorithms to tightly control the spectral efficiency of expensive licensed spectrum. Cellular technologies are without a doubt traditionally designed to make the best use of the spectrum, but it hurts power consumption of end-devices. Cellular NB-IoT and Cat-M1 have lot of optimizations in Rel 13 such as power saving mechanism (PSM) and enhanced DRX (eDRX) to put the device to sleep most of the time, but still the device has to wake up periodically to listen to the network due to the synchronous nature of Cellular IoT system.
Figure 1a. LoRaWAN Class A Operation 
Figure 1b. NB-IoT Operation 
Current Consumption of LoRaWAN Vs Cellular IoT
Every modem whether LoRAWAN or Cellular IoT has to go different states (transmit, receive, Idle and Sleep). It is the current consumption of these states that defines the power consumption calculation as we will show later.
Table 1 shows the peak current comparison between LoRaWAN and NB-IoT and it is clear that LoRaWAN is 3-5X efficient in terms of current compared to NB-IoT. Cat-M1 modems are even more power hungry but they allow more throughput for even more premium applications. Cellular IoT is indeed more optimized for high data rates.
Table 1: LoRaWAN Vs Cellular IoT Current Consumption
Note: TX current for LoRaWAN is based on Max LoRaWAN TX power =14 dBm (EU Regulations)
Airtime comparison of different states
Now, let’s look at the different airtimes between LoRaWAN and NB-IoT. The airtimes are assuming only 1 uplink payload of 50 Bytes. Typical, LoRaWAN use case like smart meters, smart lights, tracking only use a payload of 20 bytes or less. Table 5 shows the airtime comparison between different states (TX/RX/Idle) between LoRaWAN and NB-IoT and due to the synchronous nature of the Cellular IoT, NB-IoT modem spends significant time in Idle/RX states compared to LoRaWAN due to strict synchronization and scheduling requirements. We show here results for 50 bytes as this was what we found from 3GPP study item . What is very interesting and obvious in the table below is that NB-IoT modem spends lot of time in RX/Idle states due to complex protocol. And, this time increases at MCL 164 dB(which corresponds to cell-edge). The cell-edge power consumption matters most as lot of IoT devices are not mobile and if they happen to be at the cell-edge, will discharge very quickly!
Table 2: LoRaWAN Vs NB-IoT Airtime Comparison (50 Byte UL Payload)
Energy comparison of LoRaWAN and NB-IoT
Table 3 shows the energy of transmitting single 50 Byte payload and it is obvious that LoRaWAN has much lower energy consumption than NB-IoT. It should also be noted that LoRaWAN has much lower sleep energy consumption than NB-IoT due to its sleep current being 43X less than NB-IoT. Since IoT modems are sleeping most of the time, this matters a lot!
Table 3: LoRaWAN Vs Cellular IoT Energy Comparison (Single 50 Byte UL Payload)
Impact of power consumption on cell-edge nodes of NB-IoT and Cat-M1
From a recent IEEE paper  from researchers at Nokia, Telenor and Aalborg university, they calculated for rural deployment that there can be up to 4% and 17% devices in outage for deep-indoor (see Fig. 2). These devices usually are in the basement (For ex. deep indoor gas meters in basements) and suffer additional 30 dB indoor penetration loss. However, for dense urban environments, there can be more users in outage. Typically, cell edge coverage can be used by deployment of additional small cells but that revenue from 5-10% users might not justify the investment.
Fig. 3 shows the power consumption of these cell-edge users  for different IoT application scenarios from a study conducted by the same researchers and it is very clear that for cell-edge users, the power consumption of NB-IoT grows dramatically and can be orders of magnitude higher than LTE Cat-M1. In these scenarios, it might not be cost-effective for an operator to deploy small cells, but rather use LoRaWAN to extend its coverage for cell-edge users using LoRaWAN pico-cells at much lower cost. Operators can even use LTE Cat-M1 as a backhaul for LoRaWAN gateway as shown in Fig. 4 thus offering a very optimized way to combine LoRaWAN and Cellular IoT.
Figure 2: Coverage Analysis of Cellular IoT (NB-IoT, Cat-M1) 
Figure 3: Average device power consumption per day for UEs with MCL above 150 dB. (Rural scenario) 
Figure 4: Cellular IoT + LoRaWAN complementary deployment to address coverage problem for Cell-Edge users
Battery lifetime comparison for LoRaWAN and NB-IoT
Finally, in Fig. 5, we show the battery lifetime for LoRaWAN and NB-IoT for nodes located near, middle and furthest from the cell for transmitting 50 byte payloads with different frequency and LoRaWAN offers 3-5X better power efficiency compared to NB-IoT. We assume a battery of 5Wh assuming only uplink traffic. The nodes that get worst affected on both the technologies are on the cell-edge and these are for ex. deep indoor water/electricity meters (with additional 30 dB penetration loss) that are not mobile. The battery lifetime of nodes especially at cell-edge is very critical for several IoT applications as typically these nodes are static and it can be very costly to replace the batteries or deploy LTE small cells thus affecting the ROI negatively for several IoT applications.
Figure 5: Battery lifetime comparison (NB-IoT Vs LoRaWAN)(50 Byte Uplink, No Downlink)
What is most important from the above figure is the relative efficiency of LoRaWAN compared to NB-IoT rather than absolute numbers!
What kind of batteries are possible for LoRaWAN and Cellular IoT?
Fig. 6 shows the impact of current on usable battery capacity. The peak current of LoRaWAN is lowest due to the lowest complexity of the chip, whereas peak current progressively increases for NB-IoT and Cat-M1. Peak current matters a lot as it impacts the battery lifetime severely. We describe briefly different types of batteries available in the market along with appropriate technology fit:
- LiPo (used in mobile phones): This type of battery is not usable for long term usage due to ~2% self discharge rate per month.
- Alkaline: It is usable but internal resistance increases towards end of lifetime (cannot accommodate high to peak current and long lifetime) and at low temperatures. This battery can be used for both LoRaWAN and Cellular IoT, but in the latter case it will drain the battery faster with high peak current.
- Lithium-Thionyl-Chloride (LTC): This battery is more expensive, has self-discharge about 3%/year (requires 2x the usable capacity for 15 years lifetime). However, Peak-current also impacts capacity. This battery can be used for both LoRaWAN and Cellular IoT
- Coin cell (Wearables): It cannot provide high peak current so it is usable only for LoRaWAN and not Cellular IoT
Figure 6: Impact of current on usable capacity
(From technical specification of ER14505M Lithium-thionyl Chloride Spiral Battery)
So, what is the best solution?
- There is no right or wrong answer. The decision to use the right connectivity depends exclusively on IoT use case requirements which vary widely.
- LoRaWAN is most suited for applications which require to send small payloads (< 50 Bytes) not more than 20-30 times per day. Example of such applications are smart meters, smart lights, waste container monitoring and so on and require extremely low cost, low power and very small infrequent payloads.
- NB-IoT is more suited for applications that demand higher throughput compared to LoRaWAN at the expense of higher power.
- Then, we have LTE-Cat M1 which has more capacity to carry higher payloads, but at the expense of higher power consumption.
The article was published on www.amihotechnology.com as as a guest blog post with a permission from the author Rohit Gupta
 Sara R4-Series Data sheet, LTE Cat-M1 / NB1 modules. https://www.u-blox.com/sites/default/files/SARA-R4_DataSheet_%28UBX-16024152%29.pdf
AMIHO Technology are RF experts from Cambridge, UK. We specialise in wireless M-Bus, LoRa and LoRaWAN and we develop wireless communications technology such as high-performance RF modems and unique software stacks. Our core competence is product development which involves providing full integration support.
Follow these links to find out more about:
AMIHO’s RF hadrware, software licences and IP have been installed in more than a quarter of a million devices to date
#SmartMetering #SmartSensing #HomeAutomation and other #IoT applications
Have you got a wireless project in mind? AMIHO will be more than happy to clarify your concerns and advise on the most appropriate technology for your IoT application.
Get in touch with our team to explore how we could help with your ‘wireless’ ideas. We love tech conversations. Please fill the contact form below to send us a message or contact us via firstname.lastname@example.org