Besides, heating with electricity is very inefficient.
Unless you use a heat pump instead of a basic electric heater, which is what people do in Japan. A heat pump that transport 4 units of energy for 1 unit of electrical energy consumed will release 5 units of energy as heat in the home. Even if all the electricity comes from gaz with a 30% thermal to electric efficiency, that's still a 150% efficiency in the end. The best gaz heater has "only" a 90% efficiency in comparison. We should move to heat pumps for heating, it's just more efficient to move heat than produce it.
This insight is not mine BTW, I read it in the excellent "Sustainable energy - Without the hot air". It can be found for free on the web or in print. Very highly recommended.
Yes. And let's not forget that Moore's law is not related to the maximum transistor density achievable at a given time, but to the transistor density achievable at the lowest cost (see Moore's original paper, the emphasis on cost is very clear).
So far every shrink reduced costs too, but the cost reduction may stop before shrinking stops. Smaller processes would then only be used due to higher performance, but would be more expensive. As a practical example, 28nm today is still more expensive than 40nm, and people go to 28nm for performance (lower power / higher frequency or ability to integrate bigger functions), not (yet) for cost reduction. In time 28nm will become cheaper than 40nm, but strictly speaking in term of Moore's law, the reference today should still be 40nm (I'm taking TSMC as a reference BTW, YMMV with other fabs).
I expect that as we shrink further, the gap between the date a smaller process is introduced and the date it becomes cheaper than the previous node will increase, and maybe at some point cost will just increase. That will be the end of Moore's law.
There's a lot of interest riding on the continuation of Moore's law, particularly from companies that gain a competitive advantage from being among the first to get access to a smaller node. This means Intel of course, but even a lot of big fabless companies have better access to a smaller node compared to smaller fabless ones for example. The day Moore's law, then shrinking, stops, we can expect laggards in process to catch up eventually and this advantage would disappear. That would be quite a change. So let's not expect candid assessments on Moore's law status, there's just too much money involved. But among the less impacted players there's some noise starting to be heard. I noted recently the ARM CTO saying we should expect big changes, and sooner then most people expect. I wouldn't be surprised if it was related to this.
What I don't understand about telco behavior is their simultaneous enthusiasm for dragging their feet as hard as possible on infrastructure buildouts/enhancements and service pricing and for pushing dubiously mature 4GLTE!!!zOMG 433453Gigabits! based handsets that get approximately 45 seconds of battery life, which would be just enough time to run through an 'unlimited' data plan were it not horribly throttled by congested backhaul...
You're talking about different operators, having different needs.
The one operator pushing LTE today is Verizon. They have to to be competitive. They were stuck with CDMA/EVDO, which gets long in the tooth and can't provide anymore competitive throughput with the latest HSPA+. So they needed to move fast to the next step, which is LTE. Sprint in a way had the same problem, and tried to solve it with WiMAX. And now going to LTE too.
Then, there are operators with good 3G (HSPA+). They've already invested a lot in this infrastructure, and are in no hurry to move to LTE. They only do it because they're force too by competition really. In the US, ATT has to move to LTE because of Verizon high speed LTE push. But in the rest of the world a lot of operators are dragging their feet, and will deploy slowly. A bit of LTE here and there to be able to market it and make some noise about it, but not too much to limit spending.
As for LTE hit on the battery, do you remember the early days of 3G? It was the same battery shock. A technology made to last a long time is dimensioned to the limit of what's possible at introduction time, and this can be seen in the power consumption for devices.
I would amend this as follow: good spectrum is scarce independently of what governments do. The high investment costs for infrastructure and limited spectrum also naturally limit the number of players in a given area, again independently of governments actions. But by trying to milk the most money out of precious spectrum, indebted governments the world over for sure helps create an even higher barrier to entry.
This market quite naturally leads to monopolistic practices. In Europe, the EU makes some effort to balance this. There's been several fines to operators for collusive practice. Obligation to open infrastructure to virtual operators with controlled pricing (or the operator owning the network infrastructure could easily price virtual ones out of the market). Obligation to open the network to unlocked devices (since the beginning of GSM). Recently, by limiting roaming charges. But it's an endless cat and mouse, and legislator are usually not the fastest ones. But at least, in EU they try.
You already have this in the WiFi band (2.4 GHz), which is free for use. Have you ever been using a crowded WiFi hotspot? Seen how poor the experience? Well, that's how a global free for all would end up like in all dense areas.
The problem with wireless is that there is a tragedy of the commons at work: one user can improve his own experience by degrading the experience of all the others (in the form of more interference), for a global loss. The only way to optimize the overall capacity is to limit what a single user can extract. You need to have an altruistic behavior for the system to work at maximum global capacity. This is why the modem software are always locked, and have to pass strict certification by the way.
This issue is independent of the detailed technology and modulation scheme used to share the spectrum by the way. I mention this as some poster in reply to you mentions code division as a solution for frequency sharing: it's not. The issue I mention apply to CDMA, beam forming, whatever. Technological progress can push the load at which problems will start, but not enough to make the problem disappear.
A free for all solution would only work if there were more free good spectrum than needed, which is unfortunately not the case. Mobile is so convenient that whatever is available tends to be used.
Mesh is only a superior alternative for some niche applications like military and emergency network deployments. It's not a practical alternative as a cellular replacement IMHO (or some small operator would have used it successfully already).
Where mesh shines is the speed of deploying an ad-hoc network. Which explains their use for military and emergency applications.
But mesh sucks for latency: every additional hop adds some delay, there's no way around that. And backhaul is also an issue. With LTE today, a user throughput is more often limited by the eNB (name for the LTE base station) connectivity to the backbone than the air interface. This will change as the load will increase of course. But that shows that for high bandwidth wireless networks you want a fat pipe connected to each base station, and mesh doesn't help this. Relay will be introduced in LTE, but it's one hop maximum (don't want to degrade latency too much) and to help with coverage at limited cost (no need for a wire line backhaul on the relay) at the expense of the peak throughput. Because of this, even a managed mesh system in licensed spectrum would not fly.
Mesh is well understood today, it has its applications, but it's not the future of mainstream broadband wireless due to technical reasons.
An old prototype from 2003 handling low bands. Nothing commercial seem to have followed.
Yes, for low frequency systems you could do direct sampling. With a big badass A/D converter you just sample a huge band, filter out in software (DSP) the channel of interest and decode it also in software. The problem with this approach is that it's not necessarily the most efficient (many DVB-T systems still use a RF) although this may change, and doesn't address the part I mentioned about filtering.
Because in real life, you can have adjacent systems transmitting at high power while you receive a low power. This is called the near/far effect. For example, a close GSM phone transmitting at 900 MHz while you try to receive a far base station in a close band. You will have a HUGE power difference. There are only two ways to deal with this:
1) Use a RF filter to extract all power but the channel of interest. Then you're not blinded by adjacent interferers anymore and can use a reasonable A/D with a lower dynamic;
2) Have a very high dynamic on the A/D converter. The problem there is that there's a trade-off between speed (and how much band you can handle) and dynamic. There's no magic converter that can offer both at reasonable price and power consumption for mobile devices, if at all.
That's why we're still living with RF front-ends which are band specific, and SDR is in most cases restricted to the baseband processing.
That won't prevent people from dreaming of a universal wireless device anyway, because it's a very sexy idea. It's been a very sexy idea for more than 20 years actually. It's made significant progress. But plenty ignored the fine details to their detriment. My point here by the way is not to ridicule the idea or kill the dream. Just to instill some doze of hard nosed practical sense.
For a truly universal practical radio modem, we would need programmable filters and wide-band or programmable power amplifiers, with acceptable cost / size / power consumption. I'm sorry to say I don't know of anything practical there, not even on the radar. But with LTE huge amount of bands (40+, and increasing) there would definitely be a use for that. So we'll see, the big carrot may lead to a breakthrough one day?
The test chip [...] links to three RF chips for the different networks.
So it doesn't even have a wide band RF, which is something available nowadays (the article dates back from 2007, which can explains this). So expect different network specific RF FEs too. It's really exactly as I described: SDR here purely refers to the digital processing part.
There are already chips that do direct-if conversion for DVB-T.. It's not a generic chip they use but a specialized that will have a high-speed ADC and then have hardware that do the 'tuning' to the wanted frequency...
For VHF TV this could be doable indeed. It's low enough for a high-speed ADC, and you can have a front-end protecting the TV band too. And the power consumption may not be a problem in many use-cases (even mobile possibly: when the screen is on, a bit more power on the modem may not be so significant).
Still, this is not applicable to many domains
I should have clarified that I was talking about the device side, and particularly handsets. There, as far as I can see, it's zero IF and SoC embedded DSP cores as FPGA as not sufficiently power, size and cost efficient for a handset. On the network side the situation is very different. I have no knowledge of the RF architectures there, but indeed FPGAs are very popular.
It's not read-only, the SIM content is upgradable over the air using the SIM Toolkit system. It's a protocol allowing the SIM, which is a small computer (can be programmed in Java nowadays, see the JavaCard specs), to talk to a provisioning system in the operator backbone through the modem, independently of any support on the phone application processor.
In addition to the private user credentials, the SIM also contains some operator private information like roaming partner operators and their priority/preference order, and information driving the network selection process. Operators want to keep all this confidential. So operators that can control the full phone (common in the US) may not care about a SIM. But if they can't control the phone, because some users can buy unlocked phones on their own, then the SIM becomes useful. It allows the operator keeping all this information private as the SIM is secure. And the SIM belongs and is chosen by the operator itself, so they can trust it.
What you described is the difference between an old two stages RF architecture going from the target frequency to a base band signal through an intermediate frequency and a direct conversion / zero IF RF architecture. All recent RF chips for wireless are zero IF nowadays.
But SDR usually refers to the digital processing part. Some modem implementations use custom digital logic to do the processing. A SDR approach will use a big DSP (vector DSP even) to do the processing in software. Although typically some heavy parts like the FEC and FFT (for OFDM/OFDMA) can still be done with custom hardware for better efficiency.
In any case, the dream of a purely generic hardware is still only a dream. We can have flexible software modem (if you're not too concerned about die size and power efficiency), we have also wide band radios. But in front of that you still require a RF front-end (FE), comprising filters (not to be blinded by adjacent channels in Rx, or not to kill adjacent channels in Tx) and power amplifier in the transmission side. And there's no much flexibility there. You can have wide-band PAs, but it's limited and efficiency will suffer (so burn more power, heat more than a narrow band PA). And filters are not configurable. If you want to support many bands combinations, you end up with many different filters and a switch.
So the post is too optimistic. You may be able to toy a bit with this hardware, but don't expect making anything solid (product quality) based on that. Still for hacking and the fun / learning value, why not?
Actually, that is the step that should have been taken; software SIM
Why they're still going for a card is beyond me - but perhaps it has something to do with licensing.
It has to do with network operator control. It's not so obvious if you're in the US where phones are operator locked and the operator control the phone software, but in the rest of the world you can buy unsubsidized unlocked phones. Then there's no operator software on the phone and the operator has no control on what is running on the phone. But the operator still control what is running on the SIM card, which has its small CPU.
Using the SIM Toolkit (also called USAT, see TS 31.111 on www.3gpp.org) the operator can remotely update the SIM content to change the operator configuration (roaming list, etc.). It can also apply some filtering on what network access can be done and fetch various information on what's going on in the modem. On an unlocked phone, the SIM card contains the only embedded CPU the operator can depend upon to support its custom applications.
They also pursue this with the S40 platform. From the little that has been said so far, there is a revamp based on linux and Qt in the work. The S40 does address the low cost market.
But good luck to them here. If you've been to China / Taiwan and have seen what they can do in the low cost area on Android (or even without for even cheaper feature phones), there will be some fierce competition there. And the most dynamic low cost chipmakers (Mediatek, MStar and the like) are there too, with very cheap integrated AP/modem silicon provided with everything included (they even provide the production test software and process). Nokia used to depend on Infineon for low cost modem, but they've been bought by Intel and Intel has not a history to be attracted by razor thin margins.
With all the new competitors in the east low margin is not a comfortable place for Western companies who need significant margins, even if there's a lot of money to be made in the end.
Further, if they go with Android they're probably looking at legal issues with Microsoft and Apple, without any help from Google, just like every other Android manufacturer.
Do you realize the massive patent portfolio Nokia has? Apple went after them, and if my short term memory is correct it ended up with Apple having to pay Nokia (can't be bothered searching for a reference). If there's one company who do not need any patent protection, it's Nokia. Patents were not a factor in the choice.
The big factor is that they believed they would have an easier time being a leader in the WP ecosystem, and that it would be a positive differentiation vs. Android. Any money from MS is a nice sweetener, but if it drove their decision then they were nuts: it's only a small part compared to expected sales.
But in the end, they still have to compete with the Android ecosystem on price and features, and WP is not a positive differentiation at this stage for most. For now, it's a flop and it would take a lot of faith to believe it can get much better quickly. Nokia said they want to refocus on low cost WP phones now, but with all the Chinese and Taiwanese vendors targeting low cost with Android and extremely dynamic with 2G/3G/AP integrated silicon (not all markets care about LTE yet) and a large experience of extremely cost optimized designs, good luck to them. I'd put my money on the East for low cost.
I'm quite pessimistic on Nokia strategy, and believe they would have had a better time differentiating on an Android base with superior hardware, camera and possibly a hybrid Meego / Dalvik system --- add on top of Android, but still ride a very dynamic ecosystem. But we'll see. Things won't be able to last for too long as it is with some big change happening anyway. As sideliners we can enjoy the drama, but let's have some thoughts for the Nokia employees (not the managers who killed the company with silly internal bickering between Symbian and Meego and poor execution, but the ones who delivered so many great products and innovations in the mobile space).
There will likely be some time until something like IBM Watson can be applied to law in a large scale, but when this day happen it will be a bad days for junior positions in the area. And it'll happen.
Not exactly. NYSE is today NYSE-Euronext, this merger has been closed in 2007. Euronext itself was the merger of the Amsterdam, Brussel and Paris stock exchanges by the way.
What has recently been canned is the acquisition of NYSE-Euronext by Deutsche Borse. In certain markets in the EU the resulting group would have had a too dominant position, so the EU competition authorities didn't authorize the deal.
Indeed, an Erlang process is not an OS thread (or process), that's the whole point. That's how it can be lighter and you can use more contexts concurrently.
Any modern OS allows you to pin down on which CPU an OS thread will run. You can even do it at the OS process level through a GUI on Windows: just open the task manager, go to the process list, select a process and right click select "Set Affinity...". Then pick the core(s) on which you allow it to run. It's that easy. This is for XP by the way, YMMV.
You seem to be under the impression that to benefit from Erlang you need to have a dedicated Erlang OS. That may have been the case earlier on, but it would be foolish nowadays. With the pining and thread priority control of any major OS that would be a pure waste of time. It's just easier and just as good to support Erlang on top of an existing OS.
Not in Erlang, it doesn't depend much on the OS as the Erlang "process" (think of it as an actor or lightweight process) is handled in user space by the runtime. On uniprocessor systems Erlang used to use a single OS thread. Now on SMP systems it uses at least as many threads as cores of course. The runtime also handles all I/Os in a non blocking way (poll / select or equivalent).
The goal here is to have a much lighter implementation for a concurrency context than what even an OS thread can provide. As a result creating / destroying Erlang processes is more efficient than OS threads, and the footprint of an Erlang process is much lower than a kernel thread (a page at the minimum for a thread). This allows better scaling in the amount of concurrent contexts than if directly using OS level threads, while providing the isolation of an OS process thanks to the language semantics.
More information in the Erlang doc and wikipedia page.
CDMA is different than this. CDMA is just a multiplexing mechanism, splitting a single channel capacity between users. Here the scheme adds new independent channels.
And CDMA has been removed from DOCSIS 3.0. It had been added in DOCSIS 2.0, then people eventually realized it was a dumb idea over cable, and then removed it. The company that had pushed it went bankrupt, but not before its share peaked and some people made a lot of money selling at the right time...
What you mention (channel bonding) is also called carrier aggregation in HSPA and LTE (LTE advanced, not the current one). It's just adding the capacity of different physical channels and treating them as one logical pipe. Very similar to Ethernet bonding, although it's more complex when you get to the details. But it has nothing to do with CDMA.
CDMA is the most hyped multiplexing technology. It's been hyped to death, so much some people think it's some form of magic. But it's not, and it's our past now. CDMA key point was that it was the first mechanism that enabled deploying cellular over a single frequency, which maximized at the time cellular capacity. This was very useful in cellular system, but it's a non issue in cable (there's no cell, duh). So CDMA over cable is a marketing/hype driven monstrosity that should never have happened (CDMA may by useful for a contention channel though). And even in cellular there are better schemes which have become practical now. All 4G system are based on OFDMA for example, with just the contention channel using some form of code multiplexing to be more robust to collisions.
Even HSPA, which is still CDMA based, went back to something closer to TDM in spirit than CDMA: there are still codes, but they're usually allocated to a single user over a short duration, and multiplexing is mostly TDM. Instead of having multiple user at the same time using different codes, which is the essence of CDMA. The HSPA way to send with more density over a shorter period of time instead of spreading the signal is more power efficient.
It doesn't invalidate Shannon, as an AC says below and as I understand it, it creates new orthogonal channels. Think MIMO spatial multiplexing.
But as the first poster said, and IIUC from how it's done, it shouldn't be robust to multipath. If true this would limit a possibly application to microwave trunking, but wouldn't help WiFi of your smartphone.
As I explained above, this is different. What you mention share the same channel, while here it adds new independent channels. A closer analogy, although it's different, would be MIMO spatial multiplexing. This said, if it doesn't work with multipath it won't change your life.
That's completely different. GPS uses CDMA, which is a way to multiplex several users on the same channel. Here it's a way to create additional independent channels. The former is sharing one channel capacity, the later is adding channels and capacity. If you want to compare this to an existing technology, it's closer in spirit to MIMO with spatial multiplexing.
But as the grand-parent remarked, and if I understand correctly, this shouldn't be robust to multipath (i.e. all the reflections that adds up at the receiver). And all practical use cases you care about as an end user must support multipath (OFDMA used in WiMAX and LTE main strength is its robustness to multipath) as they must operate in non line of sight (NLOS) conditions. So that would limit the application to line of sight (LOS) systems like microwave trunking. Possibly still useful, but not for you and me.
And by the way, although you're correct that wireless microphones are basic tech, satellites links are by no mean state of the art. Satellite is LOS, the challenge is very low signal level but the channel is easy. The state of the art is in terrestrial broadband (mostly LTE and its evolutions now) with mobility and multipath to handle with a constrained (size and power) receiver in a smartphone.
I don't need to look for leaky applications: I have acroread opened;). With many large PDFs and I rarely close it, as it's convenient to keep these documents quickly accessible. It's the top memory user (no surprise) and nothing should leak more than this really. Even that is not bad enough that I bother, or I would use another reader (acroread rendering quality is still a bit better on my set-up. I guess I'll be able to dump it with wheezy). I close it every few weeks or so and it's ok. KDE plasma is a distant second and I never bothered closing the session just to recollect memory. Kernel upgrades happen often enough to deal with this.
The workstation is a couple years old and has a Xeon with 3 memory interfaces, so it's fitted with 6 GB (3 x 2 GB). It was not high end when bought. No swap either, and plenty of services (apache, sshd, NFS...). The other apps I open and close regularly. My 5 years old laptop has 2 GB of RAM, and I don't need to care either (but I shut it down more often).
I completely understand people on older machines with smaller RAM who want to limit memory usage, and it's very good alternatives exist. But on a reasonably recent PC that is not a netbook memory usage is not so much a practical concern as an annoyance for people who like a tight ship I'd say. Which I can understand too, but it grates instead of hurt.
Of course, I could just start kde and watch memory go down the memory hole
KDE session running on a fixed workstation since about one month here, just locked at night. Memory used: just below 1 GB. And it's not even a recent KDE, it's Debian stable 4.4 version. It's anecdotal, sure, but on a fairly recent machine the "bloat" from recent DEs seems more a perceived than an actual problem.
Yes, and the summary is sensationalist (par for the course I guess) and misleading (more annoying).
The story is just that Asus preferred the Tegra3 when having to chose a platform without LTE. And when taking into account LTE, it preferred the new Qualcomm S4 combined AP and LTE solution.
Now when you say:
The Tegra3 isn't compatible with any LTE modems and won't be for several months
It can be misunderstood as an issue on the Tegra3 side, which would be unfair. There's nothing special to do to hook a LTE baseband to a T3. It's all Android and very common interfaces. The issue is more on the LTE modem offering side: the current chips are either not available (internal developments at Samsung, Moto,...) or not attractive enough. We'll have to see a new generation of LTE modems arrive to see a nice combo based on an NVidia chip, and this is a problem for NVidia for sure. But all the work will be on the modem provider side, and it's not much work to support a T3. All modem vendors nowadays are targeting Android and from the modem support software point of view there's not much difference between one AP running Android and another.
The US built the National Ignition Facility to study nuclear fusion, so Europe is building Laser Megajoule
The NIF and ITER are two different approaches to achieving viable nuclear fusion, Europe has commited the majority of its funding to the ITER approach but it'd be stupid not to have some smaller scale experiments which use the approach that NIF uses. Just as I'm sure the US has some experiments that try the ITER torus approach.
Mégajoule (MJ) is not primarily motivated by fusion as a public energy source (although it will be used for research on that too, on the side) but by military considerations. Now that live nuclear weapon testing has been banned, ensuring proper operation of nukes can only be achieved through simulation and such tests facilities as the NIF and MJ. That's really why both were created. And by the way, MJ is not a European project but a French one: nuclear deterrence is at national level only, Europe is not involved in such military stuff.
Also, it seemed to me NIF and MJ are cooperating actually (sharing parts of the design and costs), so opposing them or saying one is a copycat of the other seems silly. Not that I'm an expert on such stuff, so take with a grain of salt and double check if you care.
Slashdot Mirror
Besides, heating with electricity is very inefficient.
Unless you use a heat pump instead of a basic electric heater, which is what people do in Japan. A heat pump that transport 4 units of energy for 1 unit of electrical energy consumed will release 5 units of energy as heat in the home. Even if all the electricity comes from gaz with a 30% thermal to electric efficiency, that's still a 150% efficiency in the end. The best gaz heater has "only" a 90% efficiency in comparison. We should move to heat pumps for heating, it's just more efficient to move heat than produce it.
This insight is not mine BTW, I read it in the excellent "Sustainable energy - Without the hot air". It can be found for free on the web or in print. Very highly recommended.
Yes. And let's not forget that Moore's law is not related to the maximum transistor density achievable at a given time, but to the transistor density achievable at the lowest cost (see Moore's original paper, the emphasis on cost is very clear).
So far every shrink reduced costs too, but the cost reduction may stop before shrinking stops. Smaller processes would then only be used due to higher performance, but would be more expensive. As a practical example, 28nm today is still more expensive than 40nm, and people go to 28nm for performance (lower power / higher frequency or ability to integrate bigger functions), not (yet) for cost reduction. In time 28nm will become cheaper than 40nm, but strictly speaking in term of Moore's law, the reference today should still be 40nm (I'm taking TSMC as a reference BTW, YMMV with other fabs).
I expect that as we shrink further, the gap between the date a smaller process is introduced and the date it becomes cheaper than the previous node will increase, and maybe at some point cost will just increase. That will be the end of Moore's law.
There's a lot of interest riding on the continuation of Moore's law, particularly from companies that gain a competitive advantage from being among the first to get access to a smaller node. This means Intel of course, but even a lot of big fabless companies have better access to a smaller node compared to smaller fabless ones for example. The day Moore's law, then shrinking, stops, we can expect laggards in process to catch up eventually and this advantage would disappear. That would be quite a change. So let's not expect candid assessments on Moore's law status, there's just too much money involved. But among the less impacted players there's some noise starting to be heard. I noted recently the ARM CTO saying we should expect big changes, and sooner then most people expect. I wouldn't be surprised if it was related to this.
What I don't understand about telco behavior is their simultaneous enthusiasm for dragging their feet as hard as possible on infrastructure buildouts/enhancements and service pricing and for pushing dubiously mature 4GLTE!!!zOMG 433453Gigabits! based handsets that get approximately 45 seconds of battery life, which would be just enough time to run through an 'unlimited' data plan were it not horribly throttled by congested backhaul...
You're talking about different operators, having different needs.
The one operator pushing LTE today is Verizon. They have to to be competitive. They were stuck with CDMA/EVDO, which gets long in the tooth and can't provide anymore competitive throughput with the latest HSPA+. So they needed to move fast to the next step, which is LTE. Sprint in a way had the same problem, and tried to solve it with WiMAX. And now going to LTE too.
Then, there are operators with good 3G (HSPA+). They've already invested a lot in this infrastructure, and are in no hurry to move to LTE. They only do it because they're force too by competition really. In the US, ATT has to move to LTE because of Verizon high speed LTE push. But in the rest of the world a lot of operators are dragging their feet, and will deploy slowly. A bit of LTE here and there to be able to market it and make some noise about it, but not too much to limit spending.
As for LTE hit on the battery, do you remember the early days of 3G? It was the same battery shock. A technology made to last a long time is dimensioned to the limit of what's possible at introduction time, and this can be seen in the power consumption for devices.
I would amend this as follow: good spectrum is scarce independently of what governments do. The high investment costs for infrastructure and limited spectrum also naturally limit the number of players in a given area, again independently of governments actions. But by trying to milk the most money out of precious spectrum, indebted governments the world over for sure helps create an even higher barrier to entry.
This market quite naturally leads to monopolistic practices. In Europe, the EU makes some effort to balance this. There's been several fines to operators for collusive practice. Obligation to open infrastructure to virtual operators with controlled pricing (or the operator owning the network infrastructure could easily price virtual ones out of the market). Obligation to open the network to unlocked devices (since the beginning of GSM). Recently, by limiting roaming charges. But it's an endless cat and mouse, and legislator are usually not the fastest ones. But at least, in EU they try.
You already have this in the WiFi band (2.4 GHz), which is free for use. Have you ever been using a crowded WiFi hotspot? Seen how poor the experience? Well, that's how a global free for all would end up like in all dense areas.
The problem with wireless is that there is a tragedy of the commons at work: one user can improve his own experience by degrading the experience of all the others (in the form of more interference), for a global loss. The only way to optimize the overall capacity is to limit what a single user can extract. You need to have an altruistic behavior for the system to work at maximum global capacity. This is why the modem software are always locked, and have to pass strict certification by the way.
This issue is independent of the detailed technology and modulation scheme used to share the spectrum by the way. I mention this as some poster in reply to you mentions code division as a solution for frequency sharing: it's not. The issue I mention apply to CDMA, beam forming, whatever. Technological progress can push the load at which problems will start, but not enough to make the problem disappear.
A free for all solution would only work if there were more free good spectrum than needed, which is unfortunately not the case. Mobile is so convenient that whatever is available tends to be used.
Mesh is only a superior alternative for some niche applications like military and emergency network deployments. It's not a practical alternative as a cellular replacement IMHO (or some small operator would have used it successfully already).
Where mesh shines is the speed of deploying an ad-hoc network. Which explains their use for military and emergency applications.
But mesh sucks for latency: every additional hop adds some delay, there's no way around that. And backhaul is also an issue. With LTE today, a user throughput is more often limited by the eNB (name for the LTE base station) connectivity to the backbone than the air interface. This will change as the load will increase of course. But that shows that for high bandwidth wireless networks you want a fat pipe connected to each base station, and mesh doesn't help this. Relay will be introduced in LTE, but it's one hop maximum (don't want to degrade latency too much) and to help with coverage at limited cost (no need for a wire line backhaul on the relay) at the expense of the peak throughput. Because of this, even a managed mesh system in licensed spectrum would not fly.
Mesh is well understood today, it has its applications, but it's not the future of mainstream broadband wireless due to technical reasons.
You should probably do a bit more fact-checking before saying stuff like that..
There's nothing below that invalidate what I've said, on the contrary. So let's go for some fact checking...
http://www.linuxfordevices.com/c/a/News/Embedded-Linux-powers-first-handheld-software-radio/
An old prototype from 2003 handling low bands. Nothing commercial seem to have followed.
Yes, for low frequency systems you could do direct sampling. With a big badass A/D converter you just sample a huge band, filter out in software (DSP) the channel of interest and decode it also in software. The problem with this approach is that it's not necessarily the most efficient (many DVB-T systems still use a RF) although this may change, and doesn't address the part I mentioned about filtering.
Because in real life, you can have adjacent systems transmitting at high power while you receive a low power. This is called the near/far effect. For example, a close GSM phone transmitting at 900 MHz while you try to receive a far base station in a close band. You will have a HUGE power difference. There are only two ways to deal with this:
That's why we're still living with RF front-ends which are band specific, and SDR is in most cases restricted to the baseband processing.
That won't prevent people from dreaming of a universal wireless device anyway, because it's a very sexy idea. It's been a very sexy idea for more than 20 years actually. It's made significant progress. But plenty ignored the fine details to their detriment. My point here by the way is not to ridicule the idea or kill the dream. Just to instill some doze of hard nosed practical sense.
For a truly universal practical radio modem, we would need programmable filters and wide-band or programmable power amplifiers, with acceptable cost / size / power consumption. I'm sorry to say I don't know of anything practical there, not even on the radar. But with LTE huge amount of bands (40+, and increasing) there would definitely be a use for that. So we'll see, the big carrot may lead to a breakthrough one day?
http://www.electronicsweekly.com/Articles/17/12/2007/42781/intel-targets-wimax-with-software-radio-device.htm
Well, in this very article it says:
The test chip [...] links to three RF chips for the different networks.
So it doesn't even have a wide band RF, which is something available nowadays (the article dates back from 2007, which can explains this). So expect different network specific RF FEs too. It's really exactly as I described: SDR here purely refers to the digital processing part.
There are already chips that do direct-if conversion for DVB-T.. It's not a generic chip they use but a specialized that will have a high-speed ADC and then have hardware that do the 'tuning' to the wanted frequency...
For VHF TV this could be doable indeed. It's low enough for a high-speed ADC, and you can have a front-end protecting the TV band too. And the power consumption may not be a problem in many use-cases (even mobile possibly: when the screen is on, a bit more power on the modem may not be so significant). Still, this is not applicable to many domains
I should have clarified that I was talking about the device side, and particularly handsets. There, as far as I can see, it's zero IF and SoC embedded DSP cores as FPGA as not sufficiently power, size and cost efficient for a handset. On the network side the situation is very different. I have no knowledge of the RF architectures there, but indeed FPGAs are very popular.
It's not read-only, the SIM content is upgradable over the air using the SIM Toolkit system. It's a protocol allowing the SIM, which is a small computer (can be programmed in Java nowadays, see the JavaCard specs), to talk to a provisioning system in the operator backbone through the modem, independently of any support on the phone application processor.
In addition to the private user credentials, the SIM also contains some operator private information like roaming partner operators and their priority/preference order, and information driving the network selection process. Operators want to keep all this confidential. So operators that can control the full phone (common in the US) may not care about a SIM. But if they can't control the phone, because some users can buy unlocked phones on their own, then the SIM becomes useful. It allows the operator keeping all this information private as the SIM is secure. And the SIM belongs and is chosen by the operator itself, so they can trust it.
What you described is the difference between an old two stages RF architecture going from the target frequency to a base band signal through an intermediate frequency and a direct conversion / zero IF RF architecture. All recent RF chips for wireless are zero IF nowadays.
But SDR usually refers to the digital processing part. Some modem implementations use custom digital logic to do the processing. A SDR approach will use a big DSP (vector DSP even) to do the processing in software. Although typically some heavy parts like the FEC and FFT (for OFDM/OFDMA) can still be done with custom hardware for better efficiency.
In any case, the dream of a purely generic hardware is still only a dream. We can have flexible software modem (if you're not too concerned about die size and power efficiency), we have also wide band radios. But in front of that you still require a RF front-end (FE), comprising filters (not to be blinded by adjacent channels in Rx, or not to kill adjacent channels in Tx) and power amplifier in the transmission side. And there's no much flexibility there. You can have wide-band PAs, but it's limited and efficiency will suffer (so burn more power, heat more than a narrow band PA). And filters are not configurable. If you want to support many bands combinations, you end up with many different filters and a switch.
So the post is too optimistic. You may be able to toy a bit with this hardware, but don't expect making anything solid (product quality) based on that. Still for hacking and the fun / learning value, why not?
Actually, that is the step that should have been taken; software SIM
Why they're still going for a card is beyond me - but perhaps it has something to do with licensing.
It has to do with network operator control. It's not so obvious if you're in the US where phones are operator locked and the operator control the phone software, but in the rest of the world you can buy unsubsidized unlocked phones. Then there's no operator software on the phone and the operator has no control on what is running on the phone. But the operator still control what is running on the SIM card, which has its small CPU.
Using the SIM Toolkit (also called USAT, see TS 31.111 on www.3gpp.org) the operator can remotely update the SIM content to change the operator configuration (roaming list, etc.). It can also apply some filtering on what network access can be done and fetch various information on what's going on in the modem. On an unlocked phone, the SIM card contains the only embedded CPU the operator can depend upon to support its custom applications.
They also pursue this with the S40 platform. From the little that has been said so far, there is a revamp based on linux and Qt in the work. The S40 does address the low cost market.
But good luck to them here. If you've been to China / Taiwan and have seen what they can do in the low cost area on Android (or even without for even cheaper feature phones), there will be some fierce competition there. And the most dynamic low cost chipmakers (Mediatek, MStar and the like) are there too, with very cheap integrated AP/modem silicon provided with everything included (they even provide the production test software and process). Nokia used to depend on Infineon for low cost modem, but they've been bought by Intel and Intel has not a history to be attracted by razor thin margins.
With all the new competitors in the east low margin is not a comfortable place for Western companies who need significant margins, even if there's a lot of money to be made in the end.
Further, if they go with Android they're probably looking at legal issues with Microsoft and Apple, without any help from Google, just like every other Android manufacturer.
Do you realize the massive patent portfolio Nokia has? Apple went after them, and if my short term memory is correct it ended up with Apple having to pay Nokia (can't be bothered searching for a reference). If there's one company who do not need any patent protection, it's Nokia. Patents were not a factor in the choice.
The big factor is that they believed they would have an easier time being a leader in the WP ecosystem, and that it would be a positive differentiation vs. Android. Any money from MS is a nice sweetener, but if it drove their decision then they were nuts: it's only a small part compared to expected sales.
But in the end, they still have to compete with the Android ecosystem on price and features, and WP is not a positive differentiation at this stage for most. For now, it's a flop and it would take a lot of faith to believe it can get much better quickly. Nokia said they want to refocus on low cost WP phones now, but with all the Chinese and Taiwanese vendors targeting low cost with Android and extremely dynamic with 2G/3G/AP integrated silicon (not all markets care about LTE yet) and a large experience of extremely cost optimized designs, good luck to them. I'd put my money on the East for low cost.
I'm quite pessimistic on Nokia strategy, and believe they would have had a better time differentiating on an Android base with superior hardware, camera and possibly a hybrid Meego / Dalvik system --- add on top of Android, but still ride a very dynamic ecosystem. But we'll see. Things won't be able to last for too long as it is with some big change happening anyway. As sideliners we can enjoy the drama, but let's have some thoughts for the Nokia employees (not the managers who killed the company with silly internal bickering between Symbian and Meego and poor execution, but the ones who delivered so many great products and innovations in the mobile space).
There will likely be some time until something like IBM Watson can be applied to law in a large scale, but when this day happen it will be a bad days for junior positions in the area. And it'll happen.
Not exactly. NYSE is today NYSE-Euronext, this merger has been closed in 2007. Euronext itself was the merger of the Amsterdam, Brussel and Paris stock exchanges by the way.
What has recently been canned is the acquisition of NYSE-Euronext by Deutsche Borse. In certain markets in the EU the resulting group would have had a too dominant position, so the EU competition authorities didn't authorize the deal.
Indeed, an Erlang process is not an OS thread (or process), that's the whole point. That's how it can be lighter and you can use more contexts concurrently.
Any modern OS allows you to pin down on which CPU an OS thread will run. You can even do it at the OS process level through a GUI on Windows: just open the task manager, go to the process list, select a process and right click select "Set Affinity...". Then pick the core(s) on which you allow it to run. It's that easy. This is for XP by the way, YMMV.
You seem to be under the impression that to benefit from Erlang you need to have a dedicated Erlang OS. That may have been the case earlier on, but it would be foolish nowadays. With the pining and thread priority control of any major OS that would be a pure waste of time. It's just easier and just as good to support Erlang on top of an existing OS.
Not in Erlang, it doesn't depend much on the OS as the Erlang "process" (think of it as an actor or lightweight process) is handled in user space by the runtime. On uniprocessor systems Erlang used to use a single OS thread. Now on SMP systems it uses at least as many threads as cores of course. The runtime also handles all I/Os in a non blocking way (poll / select or equivalent).
The goal here is to have a much lighter implementation for a concurrency context than what even an OS thread can provide. As a result creating / destroying Erlang processes is more efficient than OS threads, and the footprint of an Erlang process is much lower than a kernel thread (a page at the minimum for a thread). This allows better scaling in the amount of concurrent contexts than if directly using OS level threads, while providing the isolation of an OS process thanks to the language semantics.
More information in the Erlang doc and wikipedia page.
CDMA is different than this. CDMA is just a multiplexing mechanism, splitting a single channel capacity between users. Here the scheme adds new independent channels.
And CDMA has been removed from DOCSIS 3.0. It had been added in DOCSIS 2.0, then people eventually realized it was a dumb idea over cable, and then removed it. The company that had pushed it went bankrupt, but not before its share peaked and some people made a lot of money selling at the right time...
What you mention (channel bonding) is also called carrier aggregation in HSPA and LTE (LTE advanced, not the current one). It's just adding the capacity of different physical channels and treating them as one logical pipe. Very similar to Ethernet bonding, although it's more complex when you get to the details. But it has nothing to do with CDMA.
CDMA is the most hyped multiplexing technology. It's been hyped to death, so much some people think it's some form of magic. But it's not, and it's our past now. CDMA key point was that it was the first mechanism that enabled deploying cellular over a single frequency, which maximized at the time cellular capacity. This was very useful in cellular system, but it's a non issue in cable (there's no cell, duh). So CDMA over cable is a marketing/hype driven monstrosity that should never have happened (CDMA may by useful for a contention channel though). And even in cellular there are better schemes which have become practical now. All 4G system are based on OFDMA for example, with just the contention channel using some form of code multiplexing to be more robust to collisions.
Even HSPA, which is still CDMA based, went back to something closer to TDM in spirit than CDMA: there are still codes, but they're usually allocated to a single user over a short duration, and multiplexing is mostly TDM. Instead of having multiple user at the same time using different codes, which is the essence of CDMA. The HSPA way to send with more density over a shorter period of time instead of spreading the signal is more power efficient.
It doesn't invalidate Shannon, as an AC says below and as I understand it, it creates new orthogonal channels. Think MIMO spatial multiplexing.
But as the first poster said, and IIUC from how it's done, it shouldn't be robust to multipath. If true this would limit a possibly application to microwave trunking, but wouldn't help WiFi of your smartphone.
As I explained above, this is different. What you mention share the same channel, while here it adds new independent channels. A closer analogy, although it's different, would be MIMO spatial multiplexing. This said, if it doesn't work with multipath it won't change your life.
That's completely different. GPS uses CDMA, which is a way to multiplex several users on the same channel. Here it's a way to create additional independent channels. The former is sharing one channel capacity, the later is adding channels and capacity. If you want to compare this to an existing technology, it's closer in spirit to MIMO with spatial multiplexing.
But as the grand-parent remarked, and if I understand correctly, this shouldn't be robust to multipath (i.e. all the reflections that adds up at the receiver). And all practical use cases you care about as an end user must support multipath (OFDMA used in WiMAX and LTE main strength is its robustness to multipath) as they must operate in non line of sight (NLOS) conditions. So that would limit the application to line of sight (LOS) systems like microwave trunking. Possibly still useful, but not for you and me.
And by the way, although you're correct that wireless microphones are basic tech, satellites links are by no mean state of the art. Satellite is LOS, the challenge is very low signal level but the channel is easy. The state of the art is in terrestrial broadband (mostly LTE and its evolutions now) with mobility and multipath to handle with a constrained (size and power) receiver in a smartphone.
I don't need to look for leaky applications: I have acroread opened ;). With many large PDFs and I rarely close it, as it's convenient to keep these documents quickly accessible. It's the top memory user (no surprise) and nothing should leak more than this really. Even that is not bad enough that I bother, or I would use another reader (acroread rendering quality is still a bit better on my set-up. I guess I'll be able to dump it with wheezy). I close it every few weeks or so and it's ok. KDE plasma is a distant second and I never bothered closing the session just to recollect memory. Kernel upgrades happen often enough to deal with this.
The workstation is a couple years old and has a Xeon with 3 memory interfaces, so it's fitted with 6 GB (3 x 2 GB). It was not high end when bought. No swap either, and plenty of services (apache, sshd, NFS...). The other apps I open and close regularly. My 5 years old laptop has 2 GB of RAM, and I don't need to care either (but I shut it down more often).
I completely understand people on older machines with smaller RAM who want to limit memory usage, and it's very good alternatives exist. But on a reasonably recent PC that is not a netbook memory usage is not so much a practical concern as an annoyance for people who like a tight ship I'd say. Which I can understand too, but it grates instead of hurt.
Of course, I could just start kde and watch memory go down the memory hole
KDE session running on a fixed workstation since about one month here, just locked at night. Memory used: just below 1 GB. And it's not even a recent KDE, it's Debian stable 4.4 version. It's anecdotal, sure, but on a fairly recent machine the "bloat" from recent DEs seems more a perceived than an actual problem.
The story is just that Asus preferred the Tegra3 when having to chose a platform without LTE. And when taking into account LTE, it preferred the new Qualcomm S4 combined AP and LTE solution.
Now when you say:
The Tegra3 isn't compatible with any LTE modems and won't be for several months
It can be misunderstood as an issue on the Tegra3 side, which would be unfair. There's nothing special to do to hook a LTE baseband to a T3. It's all Android and very common interfaces. The issue is more on the LTE modem offering side: the current chips are either not available (internal developments at Samsung, Moto, ...) or not attractive enough. We'll have to see a new generation of LTE modems arrive to see a nice combo based on an NVidia chip, and this is a problem for NVidia for sure. But all the work will be on the modem provider side, and it's not much work to support a T3. All modem vendors nowadays are targeting Android and from the modem support software point of view there's not much difference between one AP running Android and another.
The US built the National Ignition Facility to study nuclear fusion, so Europe is building Laser Megajoule
The NIF and ITER are two different approaches to achieving viable nuclear fusion, Europe has commited the majority of its funding to the ITER approach but it'd be stupid not to have some smaller scale experiments which use the approach that NIF uses. Just as I'm sure the US has some experiments that try the ITER torus approach.
Mégajoule (MJ) is not primarily motivated by fusion as a public energy source (although it will be used for research on that too, on the side) but by military considerations. Now that live nuclear weapon testing has been banned, ensuring proper operation of nukes can only be achieved through simulation and such tests facilities as the NIF and MJ. That's really why both were created. And by the way, MJ is not a European project but a French one: nuclear deterrence is at national level only, Europe is not involved in such military stuff.
Also, it seemed to me NIF and MJ are cooperating actually (sharing parts of the design and costs), so opposing them or saying one is a copycat of the other seems silly. Not that I'm an expert on such stuff, so take with a grain of salt and double check if you care.