500B devices coming and nothing to connect them.

IoT is already here, and we are not ready for it

The costs of IT infrastructure deployment are rising by billions of dollars every year. Yet, it might be surprising that the Internet’s growth is actually slowing down. This means that we spend more money to reach less people. This results in either unbearable costs of deployment, or higher prices for the consumer. Either way, it looks bad.

And it gets worse. Cisco defined IoT to be the moment when the number of devices exceeds the number of people on the planet. If we agree with this definition, claims that IoT is on its way don’t make sense anymore. It already happened, sometime in 2008. Still, that’s not what the fuss is all about. IoT brought the promise for a new world of possibilities: all things connected. Lamps, heart monitors, smart watches, cars, clothes, refrigerators, people. Things. Yet, industry leaders forecast over 20 billion devices by 2020. Others, still, 500 billion by 2025. With such mind-blowing numbers, many started wondering whether the infrastructure will be ready. So far, the expectations are not favorable.

More infrastructure requirements, less people adopting it, and increasing costs of deployment. It seems that throwing money at the problem is not going to solve it anymore, and a different solution is in order. The demand for cheaper solutions is increasing, and software seems like a better alternative. Mesh Networking enables devices to communicate in proximity, even without Internet. The devices connect to each other to form a local network. They then relay content for each other, increasing range and deliverability. This contrasts with the previous scenario of deploying more and more costly infrastructure. And the potential is immense.

Still, this is a transformation that cannot happen overnight. Enabling all these devices to communicate must instead be a gradual process. Why? What do we need to do? Here’s a proposed approach and the necessary requirements to fulfill along a timeline. Read on.

Interoperability

The first step is enabling devices to communicate with each other. First, two things are not capable of communicating if one uses Wi-Fi and the other uses Bluetooth. So a need exists for compatible radios. Software, though, does not provide a direct solution for this problem. Instead, we’ll later suggest an indirect approach, called bridging. Second, devices must be using the same protocol stack. This cannot happen if each vendor develops specific protocols for their own products. It would create many small ecosystems of connected things. One alternative would be for the masses to adopt an abstract interconnected solution. Yet another would imply a rigorous standardization process. There’s no clear winner for mass adoption, but IoTivity appears to lead in standardization.

Bridging

The second step extends on the interoperability problem. Consider the figure below, with three devices participating on the same network. The device on the right uses Wi-Fi. The device on the left uses Bluetooth. Still, the two of them can communicate through an intermediary proxy that uses both. Apple released the Multipeer Connectivity Framework with iOS 7. Although it solves this problem, it’s also closed to the Apple ecosystem. For this reason, it fails to meet interoperability requirements. The Hype SDK, produced by HypeLabs, implements bridging without losing on the interoperable side.

Example of bridging different radios using mesh networking. The device in the middle uses both Wi-Fi and Bluetooth, enabling the other two to communicate.

Office and Home coverage

By now some benefits of mesh networking should already be clear. Still, 1 hop does not go much further than bridging technologies. As such, the obvious next step appears be to increasing the number of hops. More hops bring more capabilities to the network, and further extend its range. The problem is that increasing by one single hop already brings a whole new set of problems. Loops start to form, and alternative paths occur. These problems incur the need for loop prevention, self-healing, and other technical stuff. It’s a new problem of its own. The good news is that 3 to 5 hops are not that much different, so the range can already get quite reasonable. Still, it could provide an ideal scenario for home and office appliances. One example could be finally getting Internet on that damn elevator.

How 2 hops can form a loop in a mesh network. Notice how removing any of the links removes the loop as well.

Event-wide coverage

Deploying such technology at large events would be its first form of large scale. Think stadiums, concerts, or conferences, with thousands of connected devices. This would provide better connectivity, while also offloading the local infrastructure. Unfortunately, this is the point of breakage for proactive routing. In proactive networks, devices keep a constant discovery process. They are ideal for real-time communications, but fail for large networks with many devices. Reactive networks contrast with this, by establishing routes on demand. In the former case, the network starts to stutter long before connecting so many devices. As such, this is, again, a new problem to solve.

Citywide coverage

In its final form, mesh networking connects smart cities. At this point, the complexity for full connectivity is huge, maybe unfeasible. Instead, the network is segregated into several smaller interconnected networks, cooperating, and collaborating. One example of such network is what a company called Veniam did in the Mesh Valley city of Porto, Portugal. There, the citywide public transportation system has Internet connectivity through a mesh network. Berlin, Leipzig, and Barcelona are also great examples. Freifunk is the initiative behind the German solution. Guifi.net appears to be preferred by the Spaniards. Both consist of creating decentralized networks, which is comparable to some kind of parallel Internet.

Conclusion

This is a problem to address in many fronts. Key players include IoTivity, HypeLabs, AllJoyn, p2pkit, Freifunk, and Guifi.net. A fast enough deployment would either mean a fast standardization process (cof, cof) or at least one of these players reaching large-scale deployment. Either way, IoT is here, it’s big promise is still on the way, and we are not ready for it. But we are getting there.