How many devices does one LoRaWAN gateway support? ... 120?
Martin Bor and Utz Roedig of Lancaster University, UK LoRaWANTM networks recently published a paper: " “Do LoRa Low-Power Wide-Area Networks Scale?”
Their short answer: "No"
Their exact words: "Do Long Range (LoRa) Low-Power Wide-Area Network (LPWAN) scale? According to our study presented in the paper current installations based on LoRaWAN do not. "
Their simulation results: A single LoRaWAN gateway can support 120 end devices, assuming each end device transmits 20 bytes every 16.7 minutes.
These conclusions are lower than what we have heard from partner deployments and what we have calculated from our simulations. Nonetheless, it still reinforces some reoccurring drawbacks with LoRaWAN deployments
The following is our simulation numbers...(also highlights the scalability issue of LoRaWAN)
Full LoRaWAN vs. Weightless-P (here)
Other LoRa issues documented by other companies:
"LoRa has about 18% efficiency. This means that 82% of packets are lost when a LoRaWAN network is fully utilized" (Link Labs)
"LoRa is not a serious protocol...huge capacity and interferences issues...single chip vendor" (Haystack)
"The capacity for downlink messages is even lower than for uplink messages, so don't waste it."(The Things Network)
~ Thanks for reading ~
Ubiik home page - (leading vendor in Weightless-P protocol/hardware development)
We're running more than a thousand nodes on a single gateway with 15 second reports with much less than 1% missed packets, so I question the simulation and the conclusion. Our simulations show that we'll start dropping more packets (exceed 1%) when we exceed 4,000 nodes on one gateway. Note this is with ADR, so nodes are using all Data Rates, and they're time synchronized so there are very few collisions. We're in the US and every Uplink report is acknowledged with current time to keep all nodes in sync.
Senior Firmware Engineer
5 年Hi,Any formula to calculate ?
DevOps and Platform Engineer | Writing stories about the humans that make software
6 年Really interesting article - but now it's more than a year old is it worth revisiting? Is there an update to this?
CEO and Founder at Digital Six Laboratories, Inc.
7 年I think there is a problem with the math here. Even if this is a single channel gateway, it is still capable of about 33kbit/sec, but lets say for range's sake, we use SF=9 and get 7kbit/sec. That means that a 20 byte packet would require about 23mSec to transmit. So, if we divide 16 minutes into 30mSec slots, you get about 43K slots. Now, assume a network utilization of only 10%, which is much lower than LoRa is capable of, and you get 4.3K devices, not 230. However, even this analysis is off because an true infrastructure LoRa gateway will support at least 6 simulataneous channels, so the # of devices is actually much higher. Jay, maybe you can explain your math. Our company uses LoRa (but not LoRaWan) for our wireless IoT devices, so we understand this stuff pretty well, but I am curious if you have a take on this that we have not considered. Of course, this comparison is not fair for other reasons as well. LoRa is capable of variable rate operation so that closer devices get higher data rates while further ones get lower data rates, which maximizes utilization. And, LoRa is far more resilient than any NB solution in the presence of interference. Lastly, no NB solution will beat LoRa for range or battery life in an apples to apples comparison; LoRa operates at negative SNR, which NB FSK (o NB anything) cannot match. If I were in your position, however, I would not be too woried about LoRaWan as a competitor... for public network applications, Cat M is going to be the 800 pound gorilla you will have to compete with. I would love to discuss this stuff with you if you like. We face some of the same issues and I would love your insight, so message me if you want.
Senior Protocol Software Developer at RapidM.
7 年Further in the paper( section 4.1) the following is stated: Each LoRa sink is able to receive for a given CF multiple signals with different SF and BW combinations. This mimics the behaviour of LoRa sink chips such as the Semtech SX1301 which can receive 8 concurrent signals as long as these signals are orthogonal (i.e. as they are using different SF or BWsettings). Two of such chips can be used in a sink node to ensure that concurrent signals on all orthogonal SF and BW settings can be received simultaneously.