If Huawei’s New IP proposal is accepted, the architecture of the internet and how devices access it will fundamentally change. With China’s backing and support from the state-run telcos China Unicom and China Mobile, Huawei is pitching to create a new internet protocol (IP). Internet Protocols are standards that all devices on the internet use to communicate with each other.
The proposal to revamp IP is being pushed for at the International Telecommunication Union (ITU), according to the Financial Times, which was first to report documents outlining the proposal in March.
China’s pitch is that the internet today is not ready to efficiently handle networks with futuristic devices and services that use a lot of data. It says that there is a need to revamp the protocols and systems that the internet relies on. In return, Huawei argues, the internet’s plumbing will be smarter and consume less power and resources, while being more secure. Sceptics of the proposal like Mozilla and ISOC have argued that the proposal is not necessary and entails huge costs, while leaving questions unanswered on what the impact on civil rights and surveillance would be.
New IP will come up for discussion at the World Telecommunication Standardization Assembly in Hyderabad next February. The assembly, which will be key for Huawei and China’s plans for New IP, was originally scheduled for November, but got delayed. The idea of a New IP was earlier briefly floated in a Huawei “vision document” made available by the ITU in 2018.
What the New IP proposal says
Here’s what New IP is all about:
Why is a new IP needed? With content like holographic communication and explosion of “physical and virtual objects” over the internet, Transmission Control Protocol/Internet Protocol (TCP/IP), which the internet relies on to function, are no longer sufficient to address the requirements of the internet of the future. For instance, a fixed IP address length (that is required by the protocol itself) is a problem for IoT devices, which might otherwise use smaller addresses to save on processing power. Satellite technology, too, shouldn’t have such limitations. As such, a New IP would enable high bandwidth at low latency with flexible network processing, and will enable communication across heterogeneous, i.e., different types of networks.
What does New IP change? Unlike current standards like IPv4 and IPv6, New IP has i) a variable IP address length (that may save power and conserve computing resources), ii) IP addresses that can distinguish between real and virtual objects (in order to centre networks around services instead of physical devices), and iii) a user-defined IP header that allows “customised functions on data packets” (and enable so-called “deterministic routing“, which can reduce latency and data loss).
Aren’t there other ways to fix these problems? The proposal doesn’t deny this, but argues that there’s more to be done. It cites IPv6 over Low Power Wireless Personal Area Networks (6LoWPAN), an internet standard working group in the Internet Engineering Task Force (IETF). This working group was formed with the goal to adapt small devices with limited processing power to the internet protocol. Huawei argues that 6LoWPAN is too complex to implement because of computational and storage resources required. It also brings up two other proposed network architectures, Named Data Networking (NDN) and MobilityFirst, but dismisses both for similar reasons.
How would a New IP network work? A network header is a piece of information that describes a data packet —what information it contains, what it is addressed to, and so on. One of Huawei’s main arguments is that fixed rules about network headers under existing IP standards are a limitation to be addressed. Under New IP, network headers would only contain one mandatory field — a “Fields Indicator” that describes what data the network header contains. While techniques like header compression exist already, Huawei argues they introduce too much complexity. It suggests some techniques and fields that it says would improve the efficiency and speed of network routing with its header system. This is the extent of New IP’s current architecture, and Huawei says on its website that it is inviting inputs to develop on this core idea.
What does a New IP network look like in the real world? Huawei illustrates its argument with a smart home security system. An in-home CCTV camera is on standby, and as soon as a network-connected motion sensor detects movement, it alerts the CCTV camera locally with a unilateral signal that has a small network header. The CCTV camera switches on, takes a picture, and sends it to the internet through a “border router” that interfaces between the home network and the public internet. The border router adds information that was not required in the local network to the transmission, and routes it through the internet to the homeowner’s phone. Huawei says that stripped down network headers will lead to savings in power and computing resources.
Source: Huawei, via Financial Times
Huawei adds that New IP would also allow “service-aware routing”, where devices communicating with each other also talk about why a certain piece of information is being transmitted, to make routing more efficient. This would reduce latency, Huawei argues.
Wouldn’t this be difficult to implement? Huawei says to conclude that while New IP may introduce new challenges in networking, existing technologies like P4 and Protocol-oblivious forwarding could help address them. The company said that the next step after gathering inputs from researchers would be to “realize a design of the NEW IP framework and protocols and validate it [sic] in real network environments.” It did not justify in detail the costs that migrating to New IP would entail.
Will New IP splinter and centralise the internet?
Huawei denied that this proposal was aimed to make government censorship and control of the internet easier, claiming that “New IP does NEITHER define governance models for the use of those technologies, NOR lead to “more centralised, top-down control of the internet”” (emphasis theirs). But that stance doesn’t really articulate how easy building authoritarian governance would be on New IP — the proposal certainly doesn’t stand in the way of creating “splinternets” that are compartmentalised and kept out of reach for the rest of the world possible.
It will also be interesting to see if India, which is hosting the conference that will determine the proposal’s fate, will bless New IP — on one hand, geopolitical considerations against China may thwart Indian enthusiasm for a Chinese-led standard; on the other, India has been on a path to nationalising its internet, and might seek systemic changes that give it greater control over the Indian internet’s future.
