Wednesday, March 07, 2018 ... /////

A chip from Nokia Bell Labs touches the Shannon limit

Around 2000, I bought several computers and printers from Hewlett-Packard and that company was naturally overrepresented in my perspective on the tech world. Well, since 2007, it's been Finnish Nokia – a previously non-existent company. I have used the Nokia 1600 dumbphone for years, before switching to Nokia Lumia 520. And I still think that even this (now extremely cheap) phone was better in many ways than my current otherwise great Xiaomi Android phone.

The purchase of the smartphone section of Nokia by Microsoft was promising but ended as a financial failure. However, the Nokia brand got resuscitated. These days, Nokia is earning 90% of its profits from equipment to transfer data over the Internet. That's because Nokia bought Alcatel-Lucent some 2+ years ago.

Well, Nokia also began to produce smartphones again – this time with Android. Well, Nokia is actually getting $10-$30 trademark royalties per smartphone from HMD, a nearby Finnish phone maker that has bought this contract to use the brand. Nokia, HMD have headquarters across the street in Finland and you should avoid the temptation to confuse HMD with HTC – the latter is Asian. ;-) At a recent MWC 2018 event, Nokia was the most talked about brand – because of 5G (discussed below) as well as new phones including a modern resuscitation of the Nokia 8110 4G "banana phone" from the Matrix movies and a visually stellar flagship Nokia 8 Sirocco based on stainless steel.

Along with Alcatel-Lucent, Nokia has also bought the Bell Labs, the wonderful New Jersey-based deeply scientific engineering lab that has produced eight Nobel prizes. Note that in 2008, The Bell Labs' theoretical physics center was formally abolished but as far as I can see, they keep on doing scientifically high-brow stuff which was another reason (along with my exposure to the products above; and the preparation of the 5G networks) why I bought a nontrivial amount of Nokia stocks and you should consider the same. (The price of the stock dropped by a factor of five since iPhones stole the fame from Nokia.)

Nokia, Ericsson, and Huawei are the three main companies that are upgrading the mobile data across the world from 4G/LTE to 5G/NR (NR stands for "New Radio") and Nokia is arguably the most advanced one among the three. Recently, Nokia successfully completed several tests and announced that the 5G networks will start to spread on semi-commercial basis already in 2018, a year before the previous expectations.

The places which will be fast including the U.S., China, and Korea – and our socialist Europe will be the slowest region in the world. On the other hand, Czechia made a special contract with Nokia and tests to introduce 5G in Czechia will start in September or earlier.

Days ago, Nokia announced a new chip that does the "modulation" of the signals in optical fibers in a clever enough way so that it approaches the Shannon theoretical limit. Shannon is a TRF reader and the Shannon entropy and limits are various expressions that include a weighted average of logarithms. Existing subsea fibers will therefore suddenly increase their capacity in a year or so.

The trick that allows Nokia to approach the theoretical limit – i.e. squeeze as much data to the fibers (per wavelength) as theoretically possible – is known as the PCS or Probabilistic Constellation Shaping. Click at the hyperlink and look at the arXiv preprints about it. One needs to fight against the noise but certain features of the modulation are more immune towards the noise (especially "low-frequency modifications") and this fact may be used – you may pump the information selectively to these noise-immune features. The adjective "probabilistic" (synonymously "adaptive") means that the parameters determining how much of the information is pushed "where" is optimized in some way.

At any rate, I am persuaded to believe that the Nokia Bell Labs employ some of the world's best experts in the Shannon-style science about the transmission of information which is a "pretty much scientific" branch of engineering. The fact that they get very close to the Shannon limit means that there won't really be any significant room for improvements. If you only cared about the efficiency with which your optical fibers transmit the information, the equipment from Nokia could very well be "the last ones".

The switch from 4G to 5G could be fun – the bandwidths are expected to increase by a factor of 100. These high capacities are sort of needed for self-driving cars and similar applications. These new networks would be much less useful if one weren't dividing the information going through the wires according to the type in various ways. One obvious property of the data that makes it "necessary to violate the neutrality" is simply the tolerable delay. Some data must arrive within milliseconds, others may be delayed up to a few seconds (e.g. streaming video that is buffered).

I think that Nokia is one of the companies that is contributing to some high-brow scientifically loaded progress – and a practically consequential progress – that is nevertheless extremely underestimated by the mainstream media and similar venues.