> We took special pains to ensure our method was lawful and the resulting work is copyright-clean under U.S. law. We did review some non-open-licensed chip documentation and code examples available for download from open file-sharing sites. None of these materials were restricted by DRM. American copyright law contains a fair-use exception that allows limited copying and examination of such materials for the purpose of understanding the ideas and functional concepts embodied in them. We believe our download and review of those materials is fair use. Should potential copyright holders disagree with our interpretation, we invite any offended parties to engage us in rational discourse.
I really hope this works out and they figure out how to get past the IP traps in a repeatable way based off of the work in China. Unfortunately, we won't know until this is tested in court and while I'm sure there is plenty of precedent, I don't think many companies will want to take the risk. Until someone does, I fear that projects like this will fail to bring in corporate supporters (like Red Hat supports the Linux kernel) that would really move both the hardware forward by leaps and bounds. If you look at all of the open hardware today, most of it is stagnant and only exists at the behest of the chip manufacturers so that they can have easily available but rather trivial reference designs (e.g. BeagleBone, PandaBoard, OpenRD, etc).
After working for a only a few years, I've developed the opinion that electrical engineering is one of those few fields where patents and other closely guarded IP demonstrably hold back innovation. When I work for big companies on mass manufactured products, 75% of the time I get my documentation from Chinese forums [1] and parts shipped in from Shenzhen a week or two before I really get a chance to engage with the manufacturer's reps. On a few occasions, I've even received better chip errata and more up to date software from the hardware pirates than from the original manufacturers! Considering how hostile the electronics industry is to their biggest customers, let alone amateurs and makers, it's a wonder that startups using cutting edge chips can launch at all.
There are a few major electronics parts markets in Shenzhen, primarily Huaqiangbei and Luohu Commercial City (I've never been, I just hire someone with EE experience to do it for me). Most of the big name IC manufacturers' parts can be found between those two. When you are manufacturing anything in large quantities, part reliability (a better word would be authenticity) and availability are major concerns that are usually dealt with by someone upstream with experience in supply chains. When negotiating contracts I am always up front about how and where all of the major components will be sourced depending on a variety of factors.
For example, if there is a tight deadline (e.g. designing part of a motherboard based on a new chipset where time to market is critical) then the first few revisions will be made up of parts sourced mostly from Shenzhen. If it's a niche contract (like defense/aerospace) then it depends on how fast I get the primary parts (e.g. rad hardened processors take a minimum of 3 months lead time so all parts are from manufacturers starting from the first revision). However, unless there is an extreme deadline or a manufacturer I loath dealing with (cough Marvell cough) the final boards sent to the client are made with manufacturer sourced parts, for which I have to start the procurement process the second I get the contract.
So if you had to design something with medium volume(say a 1-10K/year), and you couldn't get parts from the manufacturer - would you source only from shenzen ? or skip the part ?
I'm sorry I can't really give a straight answer to that question because it really depends on the context and risk-reward calculations. If the entire design revolves around that specific part and it's specialized, it's much riskier than if the part is more commoditized (e.g. power or ethernet) and there is money in the budget to risk a partial redesign based on part availability. If there is an experienced supply chain gal on the ground in China with a Rolodex of contacts, who speaks the language, and knows how to spot fakes the enterprise is much less risky than if you're doing the shopping yourself (or remotely which is a crapshoot). If it's a low margin product you might not have a choice but if it's a high margin product you probably don't want to risk the reputation. If it's a part that requires software such as a driver on the microprocessor, then you're in a whole other world of risk management dependent on manufacturer software quality and support. So on and on. Unfortunately, I don't know of any great resources that exhaustively detail this thought process (although I'm sure some are out there), although you can always look into systems engineering, which is a whole field dedicated to understanding how to manage complex engineering projects.
Electrical engineering is all about trade offs and constraints which range from physical board size to manufacturing/assembly cost to R&D deadlines to software support. I would err on the side of investing more money into proper part selection and designing from the start with part swapping in mind because unlike with software, success or failure can be very binary with little room for error. The Shenzhen markets are great because I usually have to work under extreme deadlines that don't gel well with the pace of suppliers and my deliverables are at least several steps away from full production.
When it comes to radio, though, it's not just copyright that gets into your way.
Unlike wifi/bluetooth, which happen in the license-free ISM bands (2.4 GHz), you need your entire product certified by FCC (US) and by whomever in Europe in order to legally operate in the GSM/UMTS bands; not to mention carriers won't exactly like you if you mess up their network.
For the really interesting stuff (GSM + CDMA cellular) this is a "cross that bridge when you get there" kind of problem. Before you can get on the airwaves you need to break the Qualcomm/Broadcom stranglehold on cellular basebands and if you do that (by developing the open source hardware + software implementing RF/protocols) then the licensed spectrum bit becomes tractable.
AFAIK much of the rest of the spectrum, licensing + part procurement is significantly easier and although there are some more monopolies they are nowhere near as bad as Qualcomm/Broadcom.
OsmocomBB is not the way, what they have is a patchset over closed source GSM firmware, they merely patch DSP firmware on the baseband to bend it a little.
but we already have open source GSM/LTE implementations, everything is ready from the hardware (various SDRs), thru baseband openbts, to whole network controllers like openbsc
The FCC/ETSI certification isn't too much of a barrier, being a matter mainly of money. Any manufacturer can pay to have their product tested by a third party lab and certified to conform to FCC/ETSI regulations.
The real barrier is the spectrum licensing, which is a matter for the operator rather than the manufacturer. These at the rules that say that a particular carrier payed for the exclusive use of a band of frequencies and no one else is allowed to use that spectrum.
As I understand it though, the law is reactive, in that the law is not broken by owning a capable device, but by using it. Even then, the enforcement depends on a complaint from the carrier (ie. they have to catch you), but if you get caught the penalties are severe.
I've heard, perhaps apocryphally, that binary blobs are one roadblock IC makers set up to keep rivals from going through their implementations and finding stuff to base patent-infringement lawsuits on. (Just as in software, there are dumb hardware patents too.) I wonder if this 3rd-party disclosure might open MT up to lawsuits.
Source of this seems to be in west mentality of charging what market can bear.
Texas Instruments idea of IoT chip is $20-30 cc3000 wifi solution. Chinese company on the other hand is fine repurposing USB WIFI dongle chip and selling $2 ESP8266.
Thats right 10:1 cost difference. Guess who will make and sell millions of those.
To save people a Google: there aren't any source code or even detailed specifications yet. The talk takes place tomorrow (Dec 28), so presumably more information will be available then.
They're using a CACert (a kind of community-based CA) certificate, that isn't trusted by default on most systems.
If the root certificate you're getting is from "CA Cert Signing Authority" and matches the details given on http://www.cacert.org/index.php?id=3 then you're probably not being MITM'd.
I really hope this works out and they figure out how to get past the IP traps in a repeatable way based off of the work in China. Unfortunately, we won't know until this is tested in court and while I'm sure there is plenty of precedent, I don't think many companies will want to take the risk. Until someone does, I fear that projects like this will fail to bring in corporate supporters (like Red Hat supports the Linux kernel) that would really move both the hardware forward by leaps and bounds. If you look at all of the open hardware today, most of it is stagnant and only exists at the behest of the chip manufacturers so that they can have easily available but rather trivial reference designs (e.g. BeagleBone, PandaBoard, OpenRD, etc).
After working for a only a few years, I've developed the opinion that electrical engineering is one of those few fields where patents and other closely guarded IP demonstrably hold back innovation. When I work for big companies on mass manufactured products, 75% of the time I get my documentation from Chinese forums [1] and parts shipped in from Shenzhen a week or two before I really get a chance to engage with the manufacturer's reps. On a few occasions, I've even received better chip errata and more up to date software from the hardware pirates than from the original manufacturers! Considering how hostile the electronics industry is to their biggest customers, let alone amateurs and makers, it's a wonder that startups using cutting edge chips can launch at all.
[1] http://bbs.eetop.cn is one of my go to's