I’ve kept it under wraps for a while, but hope to make this a more public project. This lowers any prospects of me making any money from it, but does make it more likely that I will see it happen.
While campaigning and thinking laterally on the issues and technologies of the NBN. I devised some interesting aspects configurations of wireless and networking devices, which could create a very fast internet for very little cost.
Such aspects alone provide little improvement on their own, but together can form an innovative commercialisable product.
Aspect 1 – Directional links are fast and low noise
Nothing very new here. They are used for backhaul in many situations, in industry, education, government and commercially. Use a directional antenna and you can focus all the electromagnetic radiation. On its own this cannot create a 10Gbps internet system to each home for a commercialisable price.
Aspect 2 – Short directional links can be made very fast
Lots of research in this domain at the moment. Think of the new BlueTooth standard, WiGig, UWB and others, all trying to help reduce wires in the home and simplify media connectivity. If rather than connecting house to local aggregation point, we connected house to house, a short link technology could be employed to create links in the order of 10Gbps between houses. But on it’s own this does not create a low latency network. With all those houses latency would add up.
Aspect 3 – Mesh network node hop latency can be driven to practically 0
When you think wireless mesh, you think of WiFi systems. Don’t. I have devised two methods for low latency switching across a mesh. Both involve routing once to establish a link (or a couple of links).
The first establishes a path, by sending an establishment packet with all the routing information, which is popped on each hop, also containing the pathid. Then subsequent data packets include the pathid to be switched (not routed) to the destination. The first method requires buffers.
The second establishes a path, similar to the first method, excepting that rather than a pathid, the node reserves timeslices, at which point the correct switching will occur, a bit like a train track. This one can potentially waste some bandwidth, particularly with guard intervals, however the first method can supplement this to send packets in even the guard intervals, and unreserved timeslices. The second method does not require buffers.
The second method is best for reserving static bandwidth, such as for a phone or video call, or for a baseload of internet connectivity, so HTTP requests are very responsive. The first is for the bursting above the minimum baseload of connectivity.
There is a third (and fourth and fifth – multi-cast and broadcast) method which can be for very large chunks of data which can simply be sent with the route, or routed on demand. Such a method might be completely eliminated though, with too much overlap with the first.
This can be implemented initially with an FPGA, and other components, such as a GPS module for accurate timing (or kalman filtered multiple quartz system). And eventually mass produced as an ASIC solution.
Aspect 4 – Total mesh bandwidth can be leveraged with distributed content
If every house has 4 links of 10Gbps, then you can see how quickly the total bandwidth of the mesh would increase. However this total bandwidth would be largely untapped if all traffic had to flow to a localised point of presence (PoP). That would be the potential bottleneck.
However one could very easily learn from P2P technologies. And the gateway at the PoP could act as a tracker for distributed content across the mesh. Each node could be capable of storing terabytes of data very cheaply. So when you go to look-up The Muppets – Bohemian Rhapsody, you start getting the content stream from YouTube, but then it’s cut after you have established a link to the content on the mesh.
There are some problems in this grand plan to work through, but it will only be a matter of time for such solutions to be found.
The first is finding the perfect short links. Research thus far has not been on developing a link specifically for our system, at the moment we would be re-purposing a link to suit our needs, which is completely viable. However to gain the best performance, one would need to initiate specific research.
The second is installation, we need to find the best form factor and installation method for each node onto houses. I anticipate that a cohesive node is the best option, all components, including the radios and antenna on the same board. Why? Because every-time you try to distribute, you need to go from our native 64bit communication paths to a Ethernet or RF over SMA etc… Gaining latency and losing speed and/or gaining noise. However, by having everything together, you increase the distance between nodes. One possible compromise could be to use waveform conduit to carry the various links closer to the edge of the house, capped with plastic to prevent spiders getting in.
The final, is a subjective problem. That is of power consumption. However this is a mute point for various reasons. For one, the node can be designed for low power consumption. Secondly, the link technologies need not be high power devices. I’ve seen some near-IR transmitters just come out (not commercially yet), which cost about a dollar each (for the whole lot), and can reach speeds of 10Gbps, and are very low power. Finally, with the falling cost of solar panels, one could incorporate them into the node (with battery), to lower installation costs.
Opportunities / Disruptions
A new paradigm in internet:
OurNet is the current name for a reason. People own their own mobile phone and pay it off on plan. With OurNet you can own your own node, and pay it off on a plan. But in addition, the node your own becomes a part of the actual network as well as the connection to it. It forms the backhaul AND the lastmile! Hence it being everyone’s net. Such a shift in ownership is sure to have some sort of impact on consumers, a new way of accessing the internet and a new paradigm of belonging/contributing.
And of course OurNet achieves fixed-line like infrastructure with wireless technology. Individuals, Businesses and datacentres could have four (and even more) redundant links, with each link able to have multiple redundant paths. This is not possible with the current (domestic) hierarchical model of the internet, where one needs to subscribe to multiple vendors to achieve redundancy costing $$$. You could reliably host your own website from your home or business, to the world!
Faster Mobile communications
With a node on every house, mobile communication can become very fast and have very low contention. Each node can be equipped with an omni-directional antenna for short range communication. In addition a beam-form directional antenna or MIMS turnable antenna can supplement or replace the omni-directional antenna allowing for very high speed, low noise links to mobile communicators.
High Precision Positioning
OurNet is made up of fixed position nodes with super high resolution timing. If this is leveraged, GPS systems can be enhanced to the millimeter opening up further opportunities for driverless cars and the like (they already work well with visual object detection and lidar, but an additional reference point can’t hurt).