Pub/Sub model could connect IoT devices without carrier networks

Three characteristics of the Internet of Things (IoT) differentiate it from industrial automation.

1. IoT devices are inexpensive.
2. IoT devices can be ubiquitously connected everyplace and anyplace.
3. IoT devices have inexpensive or zero-cost deployment.

It explains why we see so few IoT networks and why most of the industrial IoT forecasts are measurements of industrial automation that we have had for decades.

The first one, with the exception of the issue of strong security, is easy. The second two, though, in New Jersey parlance — says easy does hard.

Ubiquitous connectivity is talked about, and there is a glimmer of hope presented by Low-Power Wide-Area Networks (LPWAN) such as Senet that focus on both low-cost technology and a business model for entrepreneurial partners to deploy networks. But waiting for carriers to perfect and deploy 5G networks to build IoT solutions will delay innovators and prevent early adopters from building proof-of-concept and prototype networks essential for the iterative learning of technical methods, business cases and making financial projections of the benefits of IoT.

Sending a trained technical representative out to deploy anything is expensive. Building an organization of technical representatives, training them, and maintaining a fleet of vehicles is prohibitively expensive. Plus, there are call centers to operate and other customer-facing systems to plan.

Creating a zero-cost IoT network

An interesting paper from researchers at Georgia Tech and University of Stuttgart (pdf) explains how a publisher-subscriber (Pub/Sub) model could be used to create a ubiquitous zero-cost IoT network. The researchers report on using existing social sensing that can conscript existing devices into a network with actionable data.

In theory, emergency responders could use humans as sensor carriers, sensor operators and sensors themselves to provide situational awareness. The idea is if the software is installed on these devices, they can interconnect, creating an ad hoc network, and send useful data to a central point from which critical events can be understood.

Examples of device mesh networks

There are good use cases that prove conscripting a lot of devices to build a network is possible. For example, when rescue workers landed in Haiti in 2010 after the earthquake disaster and turned on their phones, the mobile networks crashed from contention.

Two Mitre engineers who volunteered in the rescue effort were inspired to build a prototype smartphone mesh network that used the phone’s Wi-Fi to bypass a failed carrier network called Smart Phone Ad-Hoc Networks (SPAN). They proved that as long as network participants stayed within Wi-Fi range, they could continue to make calls and message. And if one of the phones also was connected to a carrier network, all could reach cloud services. SPAN was built in 2012 when mobile platforms were immature, requiring the Mitre engineers to root the phones and install custom software.

Political protesters added mobile technology over a decade ago with instantaneous coordination, becoming smart mobs. Hong Kong student protesters used FireChat, a messaging platform built on a resilient, self-healing mesh network that could not be controlled by any central organization or shut down. It was built later on a more mature mobile platform.

Another example is the Beacon of Hope project. Inexpensive Bluetooth beacons, registered with law enforcement and non-governmental organizations (NGOs) fighting human trafficking are left in places where victims in transit can find them. When enabled, the beacons signal distress to the smartphones of volunteers running an app the relays the signal and geographic location of the victim.

Georgia Tech’s Pub/Sub architecture

With Georgia Tech’s Pub/Sub architecture, messages are sent to subscribers that have categorized a class of messages that are of interest. Interest may be in performing an actionable operation on the message or interest may mean republishing the message, relaying it to another subscriber when a communications link becomes available.

One of the examples cited by the researchers was social sensing services to learn about the transportation infrastructure. In the example, buses would have a small computer that would query the sensors on the smartphones of passengers for information such as destination so that the optimal route could be determined.

Buses within range of one another could exchange information that, when a WAN network is within range, would be uploaded to a cloud application to determine passenger counts, service variability, predicted timetables and other metrics.

Sensors at bus stops would also query the phones of waiting passengers and use Pub/Sub to relay their location either directly if a WAN connection is available or indirectly through other passengers and buses on different routes.

Another example of Pub/Sub is the Facebook Safety Check that Facebook users have seen during the recent hurricanes, natural disasters and terrorist attacks. Facebook Safety Check uses both WAN and peer connections to report subscribers’ status as OK.

Nothing comes without a price, though. The researchers propose an API for subscribing, publishing and reading sensors. For an IoT Pub/Sub model to work, IoT makers would need to standardize on an API and run software on their devices, and smartphone users would need to install an app. The Pub/Sub network would not work for low-latency applications, at least not until a ubiquitous low-latency network is deployed by carriers and IoT network operators. And, of course, there is the issue of privacy that IoT application developers would have to adopt and clearly communicate to users.

Pub/Sub could enable many IoT applications ahead of the buildout of a ubiquitous network. It would serve as a ubiquitous network infrastructure for IoT proof of concepts and prototypes. True, says easy does hard – but reaching an agreement between IoT makers and finding a value proposition for users may not be as hard as accelerating carriers’ deployment of ubiquitous 5G IoT networks.