The current model envisages primarily video traffic using packet transport. The estimated bandwidth for a user in the near future is 20 Mbps. The corresponding traffic for ubiquitous computing will be primarily machine-to-machine communications with a large number of small bursts, and it will be an order of magnitude higher. Also, in current networks, the present-day traffic paths are primarily point-to-point or point-to-multipoint. In ubiquitous computing, a fully connected node-to-node network will be needed with broadband traffic being carried simultaneously between different paths.
The access network of today is primarily based on a node communicating with a set of user devices in multi-access mode. Ubi-comp architecture will have the devices themselves acting as interconnected nodes with mesh networking. Mesh networking entails new domains and challenges in areas of self-configurability and self-healing, and needs to be highly resistant to disruption. Current networks do not generally possess such attributes. Capability to automatically discover and recognise each other is another key attribute required for ubiquitous computing. Also, the failure mode needs to follow a graceful degradation process and default to a safe mode. The network will require universal broadband wireless coverage with no gaps or dead zones.
The usable wireless spectrum will be a major limitation to meet the expected bandwidth requirements. In particular, proximity area connectivity will demand extremely high band-width. Some level of mitigation can be done by using cell splitting techniques to smaller and smaller geographical domains using even zepto or yocto cells instead of pico or femto cells. Additional proximity area capacity may require exploring optical frequencies using networks like Li-Fi.
Unique IDs for all objects in the IoT
If there are going to be billions of objects connected to the IoT, we also need some means to identify all of them uniquely. Considering the 128-bit naming facility provided by IPv6, it appears as if there will be no problem in doing so. However, Ishikawa points out that “IPv6 is only for IP networks.” Ubiquitous computing, on the other hand, is going to involve heterogeneous networks, so the unique ID issue still looms large.
“There will be IP-connected portions and non IP-connected portions, which mainly consist of very small chips with very low CPU power. And actually, the number of objects in such a non-IP connected domain may occupy the majority of the whole domain space,” explains Ishikawa.
He adds, “For example, food trace-ability is considered a great use-case for ubi-comp, but we do not expect every single milk carton to carry an IPv6 address. IPv6 is useful when the object is a network node, but milk cartons themselves don’t communicate. The reader device for QR codes, say, on milk cartons will speak IPv6. But we certainly need to assign unique numbers to each milk carton for food traceability. At ITU-T, we are investigating what to do with such requirements, and have proposed a standard called H.Idscheme (which will be renamed to H.642.1 once it becomes official) to be used for such ubi-comp environments, and made sure such coding can coexist peacefully with legacy identifier systems. Already, a similar coding scheme is used by our team in Japan and partners in Finland.”
Calm and easy computing
Does the thought of so many computers and networks make you feel dizzy? Hopefully, ubi-comp would not be that way. A lot of research is underway across the world to ensure that.
According to the YRP-UNL, what is important in realising a network protocol for ubiquitous computing is that even general users, who do not have much knowledge of computers, should be able to easily ensure the security of their equipment, confidentiality of stored information and communication, etc. Even now, there are various types of secure protocols, but most of them require operations that need technical knowledge, such as management of authentication and encryption keys, and acquisition of certificates from the certificate authority.
So the researchers at YRP-UNL are working on communication protocols that are simpler and easier-to-use by using tamper-resistant hardware. These protocols will realise a general-purpose security infrastructure that is easy-to-use, packaged, strong and stable. It will enable general users to enjoy the benefits of secure communication infrastructure.
Another challenge in ubi-comp is to truly embed or blend the computers with the environment. Overall design should be aesthetic and calm. The power, noise and other forms of energy consumed and emitted by these devices are also critical as these should not make the users uneasy. Moreover, the computers should mingle, cooperate, support and adjust with the people populating a space, and aid them in their everyday activities. Achieving such characteristics in machines is no mean task, and involves an interdisciplinary approach including biomimetics, artificial intelligence, human-machine interfaces, physics, material sciences and more.
