Wednesday, August 17, 2016

More Thoughts on Wireless


Wireless can work in amazing ways. Just consider the above. In classic cellular world we would have say 6 beams, over 360 degrees, or six 60 degree beams. Each beam would be say 20 MHz of a 40 MHz spectrum available, and each adjacent beam would have a different 20 MHz, alternating. Thus using a classic QPSK system say for even 3G we have 1 bps/Hz and in 40 MHz we have say 120 Mbps capacity. Now for 5G, we have OFDM and we have multi beam antennas. Here we have 20 beams, and 100 bps/Hz. due to OFDM and higher EIRP per beam, and we get 40 Bbps per 40 MHz!

That is only half the tale. The other half is that the data, say video, is being compressed at higher and higher amounts.

Thus as we expand capacity we are compressing content! There are of course even more ways to manipulate this process.

In a Nature article the author makes the following statement:

The most advanced commercial networks are now on 4G, which was introduced in the late 2000s to provide smartphones with broadband speeds of up to 100 megabits per second, and is now spreading fast. But to meet demand expected by the 2020s, say industry experts, wireless providers will have to start deploying fifth-generation (5G) technology that is at least 100 times faster, with top speeds measured in tens of billions of bits per second.The 5G signals will also need to be shared much more widely than is currently feasible, says Rahim Tafazolli, head of the Institute for Communication Systems at the University of Surrey in Guildford, UK. “The target is how can we support a million devices per square kilometre,” he says — enough to accommodate a burgeoning 'Internet of Things' that will range from networked household appliances to energy-control and medical-monitoring systems, and autonomous vehicles (see 'Bottleneck engineering').

Well given the simplistic example above the tools to do this are readily available. They have been known for several decades already, only now can silicon do this. The author continues:

MIMO is already used in Wi-Fi and 4G networks. But the small size of smartphones currently limits them to no more than four antennas each, and the same number on base stations. So a key goal of 5G research is to squeeze more antennas onto both. Big wireless companies have demonstrated MIMO with very high antenna counts in the lab and at trade shows. At the Mobile World Congress in Barcelona, Spain, in February, equipment-maker Ericsson ran live indoor demonstrations of a multiuser massive MIMO system, using a 512-element antenna to transmit 25 gigabits per second between a pair of terminals, one stationary and the other moving on rails. The system is one-quarter of the way to the 100-gigabit 5G target, and it transmits at 15 gigahertz, part of the high-frequency band planned for 5G. Japanese wireless operator NTT DoCoMo is working with Ericsson to test the equipment outdoors, and Korea Telecom is planning to demonstrate 5G services when South Korea hosts the next Winter Olympics, in 2018.

As noted above the MIMO function can be at the cell site not at the end user device. Thus the above argument is a straw man at best. At worst it may be a gross misrepresentation.

On comment on this article states:

Unfortunately, this article is unsound and should be withdrawn. Currently and for many years, congestion on the Internet backbone and most local broadband in the developed world is extremely rare. They are almost never a bottleneck, especially the backbone. While traffic has gone up, as noted, Moore's Law has brought down the cost of carrying bits at about the same rate. This has been established by, among others, ..... As ... notes, the only evidence of congestion in the article (except local like a convention center) is a failure of HBO to meet demand. This is almost certainly because HBO didn't buy enough capacity, not that the Internet couldn't handle the volume. I hate denigrating the work of another writer in these tough times, but this one is so misleading it should be retracted to get errors out of the public discussion

The above is in my opinion spot on. Perhaps Nature should stick to genes and molecules and leave the engineering to those who do or have done it for a living.