Why 100 Milliseconds Matters in the IoT

By Carsten Rhod Gregersen

How can designers improve their response times in the age of remote work, driverless cars and the Internet of Things?

How fast is fast enough? Well, when it comes to connected devices, the answer is just a fraction of a second. Research shows that the expected response time between user input and device action, also known as latency, is  only 0.1 second. Anything longer than this split second and users will begin to feel as though their actions were not directly causing something to happen on the device.

Internet of Things (IoT) latency, therefore, can often make or break a user interaction with a device. With the time difference between connected device success or failure so thin, let's explore what designers can do to improve their response times in the age of remote work and driverless cars.

Why Latency Matters Now
While low latency might not be top of the priority list for many vendors and designers, it is an important part of the user experience that unlocks the device's full potential. A single second is about the latency limit for the user's flow of thought to stay uninterrupted, even though the user will notice the delay.

Anything more than the maximum one second of latency results in poor user outcomes. For example, consider a pan-tilt-zoom surveillance camera. High latency between a user's smartphone app and the device translates into controls that do not correspond onscreen. Moreover, with seconds between the input and the output, the user will start to "overshoot" the target as they fail to properly register commands.

High latency like this can occur for different reasons. With cloud- or database-driven IoT solutions, intermediary servers can lag processing time by more than five seconds—far more than the ideal 0.1 second. On the other hand, cellular network connections, like 4G, can cause slow speeds due to network traffic. One way to visualize the difference between seconds and milliseconds is to experience high latency in action. This online game asks players to land a lunar module despite its thrusters suffering from high latency, and the discrepancy between user input and onscreen output is frustratingly evident.

Why Latency Matters Going Forward
Another reason  why latency is important in today's speed of innovation is the fact that device applications are only growing more complex. From the smart factory floor to the highways of tomorrow, device connections need to be as fast as possible to adequately perform mission-critical applications.

Furthermore, one must also consider the looming network evolution from 4G to 5G. The upgrade is expected to deliver impressive data speeds, but also prompt the addition of even more devices with heavier processes. There are more than 1 billion cellular IoT connections at present, and Ericsson forecasts around 5 billion connections by 2025. Whether in relation to future robotics, driverless cars or industrial monitoring scenarios, sending sizeable amounts of captured data back to the cloud is not guaranteed to be low in latency.

How to Improve Latency
The good news is that low latency is possible and does not require designers to reinvent the wheel. Rather, they need to redesign their device's storage and connection. In terms of data storage to improve latency, edge computing is key. Edge computing is a distributed computing paradigm that brings computation and data storage closer to the location where it is needed to improve response times and save bandwidth. Since the data is closer to the device, the latency is lower. This is one of the great features of 5G: the network does enable ultra-low latency via Multi-Access Edge Compute (MEC), which moves workload processing to the edge.

In terms of connection to improve latency, peer-to-peer (P2P) should be preferenced over the cloud. Unlike database-driven IoT,  P2P technology enables direct communication between two peers without a server in the middle to slow things down. It ensures super-low latency (less than 50 milliseconds), with the bonus of high privacy and secure communication. Low latency going forward is not nice to have—it is a device must. The benefits of the edge and possibilities of P2P will be key to simultaneously lowering latency times and improving user experience.

Carsten Rhod Gregersen is the CEO and founder of  Nabto, a P2P IoT connectivity provider that enables remote control of devices with secure end-to-end encryption.