Understanding Dense IoT Networks

city-skyline

June 14, 2018

Jessica Phillips

As part of MachineQ’s ongoing commitment to service existing and new customers with our robust, secure IoT platform, we recently conducted an extensive study to evaluate the ability of our enterprise network to handle high amounts of sensor data traffic in a dense urban-deployment environment.

We wanted to stress test our Network as a Service (NaaS) in order to be able to properly advise our customers on deployment capacity based on ‘real-life’ MachineQ deployments.

The study’s results prove what we had theorized and tested in our labs over the past year: LoRaWAN based IoT networks uniquely enable multiple network gateways to receive the same message and automatically determine how to prioritize data traffic providing stronger network signals and enhancing data delivery.

MachineQ, in collaboration with Semtech, emulated a full-scale dense indoor network. We used a residential neighborhood in Philadelphia primarily comprised of 3- and 4-story brick row homes. Gateways were placed in 10 homes across a ¼ square mile area, and 100 sensors were placed in different homes within that area.

Each sensor sent randomized packets throughout the study. We increased the frequency packets were sent during each phase to achieve the same traffic as 10,000 sensors sending 250,000 packets per day in Phase 1, 500,000 packets per day in Phase 2 and 1M packets per day in Phase 3.

The joint study team assessed:

  • Average packet success rate, when sending each packet once, without acknowledgement
  • LoRa’s ability to deliver overlapping packets successfully
  • Our cloud infrastructure’s ability to withstand high-volume commercial deployments

We found average success of first transmission to be 97.73%, 97.25%, and 96.23% in Phases 1, 2, and 3, respectively. In Phase 3, at 1M target daily packet volume, this equated to 25.7% of total network load. We were pleasantly surprised to find that over 1M packets in a 24-hour period represented only ¼ of our total network load capacity.

The majority of traffic was seen by at least two to three gateways. This reinforces the MachineQ team’s confidence that our NaaS has the bandwidth to process high amounts of data — and, in the event one gateway is offline, other gateways will receive messages while that gateway is being repaired.

MachineQ offers solution providers and end-users a localized NaaS that supports their specific solutions. As exemplified in this study, the MachineQ offering is capable of handling multiple solutions across a variety of use cases, all on one end-to-end NaaS.

Through redundancy, MachineQ’s dense network and software platform can ensure that mission critical data is received by multiple gateways, often on the first transmission attempt. IoT solution providers can focus on developing their next innovative solution and end-users can focus on their IoT application(s), leaving the gateway infrastructure, software scalability, reliability and security to MachineQ.

We are looking to repeat this test in other environments with different gateway configurations. What would you like to see? How would you do this test differently if performing it on your own?