Yesterday I posted a news with progress towards a symbiotic eye. Today let me point you to the capability to implant a chip on the brain to detect “thoughts” and actuate them through a robot.

The news is on Wired and it is surely worth taking a look since it represents the completion of the interaction from the brain to the machine.

The news reports of a patient that suffers from a paralyses since 1998. She has become part of a team of researcher in 2005, trying to create a link from her brain to a robot. A chip was implanted on her brain to capture the electrical waves generated by her thoughts. These waves are decoded by a computer into signals to a robot that will act according to the patient thoughts. The diagram shows the various steps involved in this feat. Never before scientist have managed to move directly from a thought to its implementation by a robot.

Clearly, it is an encounter at a middle point between Cathy and the robot. She had to learn to think in a certain way to generate waves that can be recognized by the computer, but still it is an amazing result. It will be a very long time before we will be able to read your mind, and may be this is just too scaring to be desirable. But for people like Cathy, having this kind of disabilities, technology evolution means a lot.

Progress in electronics are promising to make the bionic eye a reality by the end of this decade. Actually, it would be better to think in terms of symbiotic eye, since the electronics can become embedded in the biological eye itself and both help one another.

We have already seen electronic implants on the brain cortex to stimulate the visual area of the brain and convey images detected by a camera placed on glasses and converted into signals by a computer, able to re-create a certain degree of visual perception.
However, this approach, although getting better and better as signal processing mimics more faithfully the signals stimulating the visual cortex and contacts to the visual cortex cells gets more and more tiny nd focussed, is not going to recreate a normal vision. That is because the normal vision involves several areas in the brain and through this approach we can only get to one.

The only hope to recreate visual capability is to use the optical nerve stemming out of the retina.

However, the attempts made so far (to a certain extent successful) are limited by the need to power the retinal implant for the detection and conversion of lights into signal for the optical nerve. A variety of approaches have been tried, none completely satisfactory and that has slow progress.

Structure of the retinal implant. Every squaree" contains the sensor and three diodes to convert infrared light into power

Now a team of researchers at the Stanford University in California have succeeded in creating an implantable chip that doubles up as a power generator.  A camera on the glasses picks up the image and a computer embedded in the glasses converts it into signals modulated by an infrared signal that is beamed to the chip on the retina through the eye. The chip is made by an array of photodetectors, like the ones on a digital camera sensor and converts the infrared signal into stimulation of the optical nerve. Here is the great invention. For every pixel (photodetector) there are three diodes that convert the infrared beam into sufficient electrical power to make the processing and stimulation of the optical nerve possible. The team has managed to create a structure that can package 178 pixel per square millimeter (with each pixel containing three diodes and the processing part. For comparison, the first retinal implant was able to provide a total equivalent of 60 pixels. Here there are potentially thousands of them (a one square centimeter would accommodate over 17,000 pixels) and researchers are already at work to produce a bendable implantable chip that can follow the curvature of the retina.

We have plenty of technologies (and products) that have been developed over the years (and keep evolving) to meet specific needs and market targets.

We are now seeing, more and more, a converging of technologies and products to solve specific needs. And I just run onto an example here.

embedded by Embedded Video

vimeo Direkt

Autisms can be difficult to detect in its early stages (that is in young children) and at the University of Minnesota’s Institute of Child Development they are experimenting a new approach for early detection: use Kinect cameras, that were developed to recognize gesture for interacting with a game. What if one places a number of these cameras in a kindergarten, as part of the ambient, and have a computer to check over time how the little children move, interact with one another and with the environment to detect early signs of autisms?

The system is being presented at the International Conference on Robotics and Automation, and that is not the place where I would have expected to see people discussing autism support! Indeed, by exploiting a variety of technologies it becomes possible to solve problems in quite different areas. I really feel this is going to be what will be characterizing the next 5 years. Leveraging from diversity. As each technology/application/data is being made available in an open framework people will find new, smart and ingenious ways to use them.

There have been, over the last 30 years a number of claims that the Moore’s law was about to reach its “end of the line”. And every time a new approach to evolution of silicon manufacturing found a way to circumvent the stumbling block. And at present, as I posted in the last weeks, we can be confident that the Moore’s law at the chip level is still going strong and it looks like it will keep going through this decade.

The Moore’s law, that is about the single chip since it is tailing about the ever growing density of transistor on the silicon sliver, has the implication of leading to an ever increased processing speed and therefore it has been often applied beyond the chip, as an example to computers. One could well say that if it holds for the chip and a computer is a chip then…it holds for the computer as well. And it has been so for many years. But in the last twenty years the increase in processing has started to be a function of the number of processors running in parallel, at least for supercomputers. Then in the last 5 years we have started to see parallel processing also in desktop computers and now we are seeing them in tablets and in cell phones.

Still the Moore’s law, in terms of ever increased processing power has hold true.

It is on this front that a few voices are starting to see the end of the Moore’s law validity, and more specifically in the ever increasing processing powers of supercomputers. Take a look at the graphic:

Increase of processing power in supercomputers

Supercomputers have now reached the PFLOPs range (million of billions operations per second). In the eighties they were running at a billion operations per second and they reached the thousands billions in 1999 and the million of billions in 2008. Hence, by the end of this decade we should hit the Exascale processing (billion of billions) and in the next decade it should move towards the ZettaScale but, according to Thomas Sterling, one of the Superomputers guru, this will not happen and he is unsure if the Exascale will be reached.

These are mind boggling numbers, so one may not feel over concerned by them NOT happening! Besides, it is about supercomputers, not about the small chip. However the small chip lags behind, in terms of processing power, by about 20 years. Hence, one might extrapolate and say that Moore’s will cease to apply in the 2030 decade. Still pretty far away to become really concerned.

Besides, why would one want a chip to process zillions of operations per second? Well, this kind of questions have been asked in the past and they proved silly in hindsight. But the issue that I see is not the possibility of not being able to process ever more complex problems since the processing increase will come to an end; rather, the big issue is related to the working of our economic ecosystem.
In the last 60 years companies and economic markets have teemed on the “certainty” that technology will evolve, will become more performant and cheaper. This assumptions have shaped our economy and the “market growth”. This, I feel, is the del challenge we will need to face and there are several strong hints that this will indeed happen in this century, possibly well before 2050. And that will completely reshape our world. The silicon age will be replaced possibly by the carbon age but that will not be enough to keep the economic rules of growth we have been used in this last 60 years.

Printed electronics is now a reality, it is being used for printing RFID tag at a very low cost, for solar panels and in some special circuit building. Now researchers at the University of Chemnitz have announced a process to print a loudspeaker on paper.  Take a look:

embedded by Embedded Video

YouTube Direkt

The loudspeaker is built by overlaying several layers of print. As a support you can use normal paper, like the one of your daily newspaper. The only drawback, so far, is the need to connect the newspaper to a sound source, like an Mp3 player with a wire. But I am pretty sure that also that can be solved in the near future by embedding an antenna in the paper thus receiving the signal wirelessly (as it is done in RFID tags). What about the battery? Well, I guess one can imagine to print on the newspaper a solar panel, that should work pretty well on the beach!

Imagine listening to your newspaper, not just reading it. Hearing interviews, music…. Even the old familiar paper may become a multimedia interface and experience.

Clearly, mine are just speculation, however the reality is that technology keep progressing in two important directions: more performance and cheaper cost. And together they are changing the world.

The new memory generation: the DDR4

Samsung has started delivery of the first batch of DDR4 memory, the next generation of memories for computers. As usual, these memories will first find application in servers and only in 2014 will find a way to our laptop.

These memories work at 1.2 V, that is 20% less than DDR3, and that means less energy usage (and so less dissipation). Their speed is also better than the present generation , reaching 3.2 billion transfer per second, and that translates into 2.4 gigabits per second.

Again, we are on track with Moore’s law. Faster, cheaper, less power hungry. Let’s enjoy till it lasts!

I remember the times when I discussed if a classroom with more than 20 students was too crowded to get an effective education. Well, that was then, now is now. And now Harvard and the MIT have jointly announced the EdX program to “revolutionize” education in the world (their words).

Take a look at the announcement:

embedded by Embedded Video

YouTube Direkt

The announcement to join forces and provide an extensive set of courses taught by their best professors to be used, for free, by people all around the world, follows a first experiment that has demonstrated a strong interest. A course in electronics has been attended by 120,000 students, ranging from university students to college ones. One  ”pupil” was 80 years old.

The courses are not just nicely presented material. They include tests and ways to find out if you really understood and learnt. We are really moving towards a world where if you just want to do it, you can do it. And this is affecting also the roles we are playing. It used to be that you either were a student or a professor. Now the distinction fades away. My youngest son, 15 years old, was asked to prepare a lesson on rational numbers. I started to hand some help and then I showed him how to use iBook for preparing his lesson and organizing the material. In less than a week he along with a few classroom friends has prepared a book on the topic. And if they decide so that book can be published and become available to anybody in the world (for free or at a price, up to them to decide!). If you look at the result you wouldn’t tell it comes from a few 15 years old students! I can easily imagine millions of book being created in the coming years by students.

Barriers are crumbling as everything becomes accessible and affordable.
At the same time, many players will need to rethink their position in the blue chain since a good portion of it is fading away.How can companies charge for something that is readily available, in quantity and quality, for free? This is the big challenge most industries will have to confront themselves with.

In a previous post, we’ve started elaborating about Soft Defined Networking (SDN), which is about virtualizing network equipment and decoupling them from network management and control; not only this, a key facet of SDN is introducing API for programming network services. In principle, this could mean morphing routers into commodity (low cost) programmable boxes controlled and programmed (through API) by an outside source.

In this interesting paper (Above the Clouds: A Berkeley View of Cloud Computing), they have mentioned how two-thirds of the cost of WAN bandwidth is the cost of the high-end routers, whereas only one-third is the fiber cost. So, according to them, simpler routers built from commodity components (as SDN is planning to have) deployed in WAN, may provide costs dropping more quickly than they have had historically.

Naturally, traditional Technology Providers doesn’t fully share this view, above all about commoditizing network equipment; they are proposing another view about the emerging SDN: it could be a mean for having more interfaces to network equipment and more functionality.

In this direction, recently, Cisco has elaborated about a platform, which will be called Cisco Open Programmable Environment, providing this claimed programmability and visibility at multiple levels of the network. If OpenFlow, does this at the control and data planes, it seems that Cisco plans to expand these concepts to the rest of the network, up to the orchestration layer. Apparently this a way to use the huge amount of state information held by a network.

But there is more around. OpenStack is just an example.

OpenStack is an open source cloud project and community with broad commercial and developer support. OpenStack is currently developing two interrelated technologies: OpenStack Compute and OpenStack Object Storage. OpenStack Compute is the internal fabric of the cloud creating and managing large groups of virtual private servers and OpenStack Object Storage is software for creating redundant, scalable object storage using clusters of commodity servers to store terabytes or even petabytes of data.

Interestingly, OpenStack has a network connectivity project named Quantum (project page here). Quantum looks to provide “network connectivity as a service” between interface devices managed by other OpenStack services. Quantum itself does not talk to nodes directly: it is an application-level abstraction of networking. It requires additional software (in the form of a plug-in) and it can talk to SDN via an API.

Imagine these “paradigms” starting being exploited virally at edge of the networks…

Well, you may not like them (actually i don’t…) but they are clearly human interfaces! I am talking about the devices that researchers at Autodesk Research, based in Toronto-Canada, are testing to be embedded into our bodies. You can see a few of the right here:

Various kind of sensors and actuators designed as implant under the skin...

As you can see in the picture there are sensors to feel the pressure (imagine using your arm as an input device: tapping your finger on it activate the pressure sensor that wirelessly communicate the taps to a computer for action), to feel the presence of something (capacitance sensor), the vibration, a little speaker (place it in your hand and you won’t need a phone , just cup your hand on your hear – creepy isn’t it?), a microphone to speak at your finger, a bluetooth for local communications and and inductive charger so that when you are seated at your desk all the paraphernalia embedded under your skin recharges.

Well, I cannot say I am longing for having any of them injected under my skin but the fact that researchers are studying the technology to mke it feasible tells me that somebody sometime will try (some of) them. And, of course, it may be some sort of geek or someone having specific needs.

I cannot imagine myself morphing into a sort of cyborg, but again if I think back of the very first cell phones on the market 30 years ago I remember saying I will never get one of those bulky thing to phone. Why should I? If I need to make a call I’ll just use a telephone boot!

Complex Systems, despite their names, are composed by simple components. Complexity arises from the local interactions. One of the core questions for engineering and exploiting the extraordinary properties of complex systems is how to define and use simple local rules to generate higher levels of organizations. But we must answer also the questions if this “emergence” higher levels are stable or instable, temporary or permanent and so on.

Complexity emergence is a fascinating subject of study in mathematics, physics, biology, social sciences…but not only, it is also very often considered when studying future networks as complex systems of communication, processing and networking resources.

Interestingly, this paper elaborates a view of emergence of life by analyzing the mathematical properties of autocatalytic sets (collections of molecules which catalyze each other’s reactions, helping to bring each other into existence).

Simple example of autocatalysis

Autocatalytic sets have a complex structure of their own: imagine a system of multiple loops and chains, loops within loops, mutual cross-feed relationships connecting them, inhibitory connections, preferential reactions given different substrate concentrations…like an ecosystem! Paper argues that “ self-sustaining, functionally closed structures can arise at a higher level (an autocatalytic set of autocatalytic sets), i.e., true emergence“.

What makes the approach so interesting is that the mathematics does not depend on the nature of chemistry, i.e. it is substrate independent. So the building blocks in an autocatalytic set need not be molecules at all but any units that can manipulate other units. These units can be complex entities in themselves.

Also economy is essentially the process of transforming raw materials into products that themselves facilitate further transformation of raw materials and so on. So they argue that “Perhaps we can also view the economy as an (emergent) autocatalytic set, exhibiting some sort of functional closure“.

Could it be that this theory of autocatalytic sets can provide a new mathematical approach for modeling future networks ecosystems ?