Device-Based Mesh Network Seen as Step beyond Cellular

Rajeev Krishnamoorthy, CTO, Scintera Networks

Rajeev Krishnamoorthy, CTO, Scintera Networks


Smartphone Computing Power Said to Be Sufficient to Support Shared Peer-to-Peer Air Links

Does the world really still need cellular infrastructure for very high-speed robust wireless communications? Not everybody thinks so, including two senior engineers in the field who suggest there’s enough computing power in today’s smartphones to support a wireless mesh communications system that would operate on a peer-to-peer basis where each phone is a node in a constantly shifting crowd-based network.

Ken Pyle, video editor for ScreenPlays and managing editor of Viodi.tv, recently caught up with the authors of technical papers describing how such a network could be built. They are Devabhaktuni “Sri” Srikrishna, CTO of ABB Tropos Networks, which builds metro-scale Wi-Fi systems based on mesh technology, and Rajeev Krishnamoorthy, CTO of Scintera Networks, Inc., a supplier of analog signal processing equipment. Their plan, originally conceived as an answer to authoritarian governments’ control over communications among dissidents, is described in two papers: “How to Make Twitter Available in North Korea,” “ and Social Mesh: Can Networks of Smartphones Ensure Public Access to Twitter During an Attack?”

These are not entirely new ideas, but these experts say they now are doable in the mass market owing to the prevalence of handsets that could support such a network. Commercial applications could well begin with efforts to enhance existing cellular network capacity through use of unlicensed spectrum, but Srikrishna and Krishnamoorhy see much bigger opportunities ahead. An edited version of their conversation with Pyle follows.

Ken Pyle – In 2014 we’re seeing a lot of changes with spectrum and the potential for changes. I ran across a presentation from your paper on the idea of social mesh networking. Why don’t you tell us what that means?

Devabhaktuni Srikrishna – The way we defined it was an indestructible wireless broadband network that everyone can use.

Pyle – And in your background as CTO of Tropos Networks you know a lot about mesh networking.

Srikrishna – That’s right. We built infrastructure mesh networks in the unlicensed spectrum that are deployed in over a thousand cities around the world and some very large regions as well.

Pyle – And Rajeev you come at this from a different angle but also a wireless networking angle.

Rajeev Krishnamoorthy – I had a company which did very wideband communications over unlicensed spectrum. Over there the whole idea was you don’t have to have much regulation, and you can actually construct an extremely robust communications infrastructure. One of the ideas behind social mesh is in fact using spectrum, sometimes beyond the government purview, to establish sort of a democratic communication between thousands of people using this type of device [smartphone] without infrastructure.

Pyle – So really using these devices, but instead of going back to the cell towers, going directly from one to another.

Krishnamoorthy – That’s right. The whole point, and Sri can elaborate on this even more, is that’s the way the indestructible nature of the network, how the Internet itself started. That’s something that has somewhat been forgotten, but being revived by what we’re doing right now without infrastructure.

Srikrishna – That’s correct. That was definitely my inspiration for designing this. The Internet, the idea of packet networking came from Paul Baran, who lived here for many years, and he came up with this idea of a network where even if the Russians attack us with a nuclear weapon and one part of the network was broken, the other parts would still work. So it was resilient, and that was the original idea. That led to the Internet, which led to many more applications.

But even before the Internet there was this idea of a wireless mesh network that was designed here in the Bay Area for making wireless boxes that talked to each other so that they could pass messages back and forth. This was designed for the military in the 1970s. Wireless mesh has had several iterations over the last three decades. Tropos Networks is one of the many different wireless mesh networks.

We asked ourselves, can we design a wireless broadband network that anybody can use but cannot be destroyed or compromised by anyone? To do that, one of the things we realized is you can’t rely on infrastructure, because infrastructure can be destroyed or taken down by a hostile government or people trying to prevent freedom of speech. So we realized it has to rely on the smartphone or something that’s in the hands of people in order for it to work. And it can’t just be in a specific narrow band. It has to be over a large spectrum so that it cannot be jammed. And there are other types of counter measures we realized an attacker could try to use. So we had to design a protocol and design methods that prevent the attackers from being able to disrupt or destroy the network.

Pyle – This was written as a white paper about a year or so ago, and really you guys have been evangelizing it since then. What has been the reaction thus far?

Krishnamoorthy – We’ve done it in two ways actually. Initially when we wrote the paper, it was all about how you promote communications in oppressive countries like North Korea. It wasn’t publishable in technical journals because it was too political. And it wasn’t publishable in political journals because it was too technical. So we split it and got one part published in the FAS (Foreign Affairs Society) publication and the other one in IEEE Communications.

The original idea was really how do you make communications possible in countries where dissidents are trying to get out from under undemocratic government. But since then there have been lots of interesting discussions on how do you do disaster recovery, where there’s not  necessarily a jamming concern, but there’s a lack of infrastructure. If a place is devastated you don’t have time to build base stations.

Then there’s even the concept that you have the infrastructure-based communications systems that generally take a lot of capital and are slow to grow out. In futuristic applications of cellular communications, can you just have extremely wideband peer-to-peer protocols that allow general communications and not just rely on cellular companies to establish everything? And for that you really need a system  that  both delivers broadband, that delivers large range and really allows you to use your neighbors in order to establish a very fast linkage mechanism.

This social mesh network delivers all of those. And so the discussion really has been on all these tangents.

Pyle – That’s what kind of piqued my interest, the idea that right now there are plans to spend something like $7 to $10 billion on the FirstNet initiative. You have the auctioning of spectrum. And it just seems that the technology, where it has come to with these devices, with your approaches to the social mesh network – all of this kind of mixes together, and policy makers should probably be looking at this as a way to move forward.

Krishnamoorthy – Right, and I think the couple of papers that we’ve published, and I think Sri  has since then published even a couple more, are good basics for people to read up on in much more detail than we have time to get into here in order to understand what’s behind the system – how you use it to transform wireless communications to make it more appropriate for the next generation rather than completely rely on the existing model of bought spectrum, narrowband and cellular infrastructure.

Srikrishna – Just to add to that, I think the future of cellular communications involves crowd sourcing and a crowd-based approach, because in the last decade we’ve seen the world has gone from a very limited number of devices with very low computational capability to a very huge number of Android, iPhone and other types of smartphones that are just sitting out there with humungous super computer capabilities that can forward packets back and forth. Wireless networks have not yet adapted to take advantage of that change.

In some of the follow-on work we did to this paper, we found we could effectively double the capacity of cellular networks today and improve the reliability from like two 9s to much, much larger than that. It would allow communications during disasters even with cellular networks. So there’s a lot of opportunity to take advantage of this crowd-based approach.

Pyle – That raises the question, in some scenarios of your concept, could you use the existing radio hardware that’s already in these devices?

Srikrishna – Absolutely. There are ways of using the Wi-Fi interface that’s already here to connect. There are companies that are doing this. Open Garden is one company that has millions of users that are doing this, using Wi-Fi.

Part of the lessons of the papers we’re writing is that if you were to design a radio within known technology that was much more well suited to this purpose, you could get some enormous benefits from the existing cellular network in terms of improving the capacity, improving the reach, improving the ability to connect during difficult times, as well as to be able to connect even when a government or an entity is trying to stop you.

Krishnamoorthy – This system can be layered on in a number of ways. You can start off with using existing smart phones in order just to get better capacity, better throughput, better reliability in an existing network. And then, we continue to talk about cellular communications, which means there are base stations with cells around them. But eventually you get to a point where you can just have peer-to-peer communications with certain egress nodes; you don’t necessarily need base stations.

And the third problem, which is really the Holy Grail, how do you take care of the spectrum allocation problems? That’s a completely manmade issue in terms of the allocation procedure. It’s a nightmare because it’s a scarce resource. So how do you get it so that either you make it more efficient in terms of getting incumbents out or you simply remove the need of allocating spectrum and make everything extremely wideband so that users themselves, democratically, take whatever spectrum they have?

There’s an etiquette for doing that. And that would solve the basis of one of the points of our papers, which was having wideband spectrum allocations that you don’t need to allocate piecemeal geographically, but rather allow them to use all the bandwidth they need and have an etiquette for sharing.

Pyle – And that’s only possible because of digital and things like direct sequence.

Krishnamoorthy – That’s exactly right. And part of what that enables us to do is to have spectrum flexibility between supporting high data rates and higher range. There’s always a tension. In the beginning you want to have high range because you may have very few users around you. But once you get to the point where you have a lot of neighbors you want very high capacity.

How do you fluidly move from one end to the other? The manner which we describe in our publications allows you to have both of those, not at the same time, but when you need to you can have extremely high data rates for video applications, etc. At other times when there are sparse surroundings, it allows you to have very high range and get to very long distances.

Pyle – So you’ve looked at both licensed and unlicensed frequencies. Is there a kind of ideal range as far as a band that this would work in?

Krishnamoorthy – The difference between licensed and unlicensed bands is that in licensed bands you are guaranteed some connect reliability and throughput. That’s why cellular companies pay billions of dollars to get narrow slices of spectrum, because they own it and they can do whatever. But it’s sort of a primitive way of assuring reliability.

A better way is to assure very high reliability and very high quality of service without necessarily selling spectrum. But, given all that, you’ll continue to have some licensed spectrum for sure. But our preference is to have predominantly unlicensed spectrum, because then the marketplace really evolves to have much more creative architectures and to have things like very high bandwidth, very high range, very high reliability, etc.

In the original premise of our paper, since this was in countries where you are operating beyond the law, the whole concept of licensed spectrum is meaningless anyway. North Korea’s main problem is not that you’re violating spectrum regulations; you’re a dissident. In that context it’s clearly unlicensed and unregulated spectrum.

Pyle – What sort of bandwidth would be ideal? I know you’ve talked about 500 MHz.

Krishnamoorthy – From a conceptual point of view, the more the better. But from a practical point of view, because you have to have radios that are actually capable of delivering good linearity, and you have to have analog-to-digital converters that can sample, 500 MHz to 1 GHz is completely feasible in today’s technology. You can have more, but probably that’s a good quantum of spectrum to start with.

Then you see how you can add on. In many cases you can have things like channel bonding. So you can start out with some amount of spectrum and you can always bond more channels. You clearly want several hundred megahertz to start off with if you want to apply some of these interesting broadband techniques.

And, btw, this is quite separate from some of the other infrastructure-free mesh concepts, which still operate over narrowband channels. You have a lot of benefits there, and having the broadband component just adds even more benefits.

Pyle – In some respects this seems like ham radio with packet radio.

Krishnamoorthy – The very first wireless network that operated with a packet protocol is ALOHAnet, invented by someone in Hawaii. That was completely peer to peer, and it was sending across different islands. So, really, going back in history, the very first functioning network and protocol that existed was mesh, was packet, and was completely peer to peer. So, in a sense, things come back.

Pyle – I guess the big difference is now you have software-defined radios that you can put in these little devices. In one sense this has been proved in theory and in practice with ALOHAnet and so forth, but do you see any opportunities for piloting this anywhere?

Srikrishna – Yes, at a pilot level, we would need to do it in some place where the spectrum is available to use, and I think unlicensed spectrum is probably the most likely candidate. And I think the other thing that’s happening is we have this spectrum crunch or capacity crunch in mobile networks where mobile usage is going up so much, that that might actually bring about the need for this architecture to enhance capacity of all the phones we’re using. So there’s a very real possibility that we might start seeing this on our cellular networks even sooner than anywhere else.

But in terms of piloting the network to enable freedom of speech throughout the world, that’s something we would like to see happen. That won’t happen based on the cellular technology that’s being developed today. It needs a special effort.

The other part of this is there are some programs that are sponsored by the State Department to enable certain small groups of people like dissidents or whoever to communicate. And one of the things we realize is those can also be targeted by a government or someone who is trying stop them from communicating, whereas if the network is everywhere, if everybody has the network, there’s no way to target it. That’s a fundamental leap we have to make as a society to enable everyone to communicate all the time.

Krishnamoorthy – I would say there are three scenarios I can think of that would result in reasonable pilots. One of them is the State Department, which, as Sri was saying, is clearly a way to see whether we can target dissidents and give them access to it. The second would be FEMA or some analogous agency with extreme disaster recovery, because I think with a very low cost you could set up a very nice peer-to-peer protocol communication system, and it would really help in those disaster situations.

The third would be to build a test bed, probably in an academic environment. But my inclination would be to go toward one of the first two, because when you have a targeted focus and an application it helps you resolve what you really need to build and in what way. So you know the State Department or something like FEMA would be extremely useful to build pilots at very low cost.

Pyle – Is this the type of thing where you could literally use existing phones in a pilot situation?

Krishnamoorthy – Existing phones, or you could do some variation. There are lots of communications chips out there. It’s very straight forward. I’m not sure it’s in our papers, but we actually did a bit of materials analysis as to what it would take to build this type of phone, and these are fairly inexpensive. So the technology part and development is very straight forward to do once the motivation is there to actually have an application where we can say this is the kind of demonstration system or pilot we want to do.

Pyle – What about corporations? Have you gotten feedback from some of the larger companies? I could see there would be some natural allies for an approach like this.

Krishnamoorthy – We haven’t done a whole lot. Frankly, after we wrote the paper, during the Arab Spring a couple of years ago when all these people were getting shut down, we thought wouldn’t it be nice to design a system that they could use robustly. And so we came up with this, and then it started getting some interest. But we both have other things to do, so we’ve been busy.

Pyle – That’s one of the things I find refreshing, you’re not really here pushing an agenda.

Krishnamoorthy – We haven’t really pushed it much, but every time we have mentioned it, the interest is phenomenal. So I think it’s worth pursuing a little farther, and we will probably start looking at avenues to push this – from a government point of view and also from a corporate point of view. In other words, what are the commercial applications of this? I can envision a whole range of possibilities that need to be explored. It really needs to be followed up a little more.

Pyle – Well, hopefully this will help in that regard. I really appreciate you both giving us an update on this.