Opening the discussion, Leo McHugh, VP, instrumentation, aerospace and defence with Analog Devices, suggested that customer expectations were changing.
“In the past, our customers would have the skills in-house to put systems together; we just supplied the components. That has long gone. Today, they are taking a sub-system approach and we are supplying full system modules.
“Crucially, time is the key differentiator, especially as the pace of change accelerates. Customers are no longer concerned about what’s inside the box – it just needs to work and do what is required.”
Arun Ghosh, AT&T’s director of advanced wireless technologies, agreed, describing time as a ‘luxury’. But, from his perspective, he said he was benefitting from this trend.
“When it comes to 5G, the pressure on timing is significant. We have just two and half years to ramp up to 5G, compared to six years with 4G/LTE. People don’t want components, they want integrated systems and those systems are crucial to delivering 5G to that timetable.”
Another area of debate centred on data and how it was being generated and used.
According to Stan Schneider, CEO of Real Time , the tests associated with technologies such as 5G, autonomous vehicles and the industrial IoT were generating masses of data.
“Using analytics, companies are generating vast amounts of data, but the question is whether that data is useful and are we able to use it? We need to start differentiating between data and useful data. Look at an autonomous vehicle, for example’; it will be generating data from numerous sources, such as cameras, LIDAR and radars; even tracking the actions of the driver. Try and capture all of that all the time and you’ll be generating a lot of useless stuff.”
Schneider talked about the benefits of a data bus, as opposed to a data base.
“Unlike the latter, a data bus will only identify and use the data it needs to manage a particular situation. Useful data is more important than the volume of data and, ideally, we should be able to figure out automatically what data is needed and, ideally, do that locally.”
Starkloff said he knew of one customer that generated so much test data that there was insufficient bandwidth to send it over the network. “It’s faster for them to get their data back to base via hard disks loaded onto an aircraft,” he said.
“5G channel sounding techniques need to collect vast amounts of data with thousands of parameters. We need to be able to optimise machine learning and data mining if we are going to develop the models to emulate, model and simulate 5G networks,” Ghosh said.
Whatever the challenges – and the panel talked about the destructive dislocation often associated with new technology – the contributors agreed that there were huge opportunities in the future and the levels of connectivity being created were going to change the world radically.
