Artwork from the Print edition by Tanmayee Deshprabhu
This article was originally published in the Oxford Scientist’s Print edition, Networks, in Trinity Term 2022. You can read more of our Print articles here.
What is the first thing that comes to mind when someone says “future technology”? Flying cars, robotic butlers, and artificially intelligent cities are staples of science fiction, but some ideas within these fantasy worlds may be more plausible than they initially seem. It is far from uncommon for science fiction from the 19th and 20th centuries to have predicted our present – Jules Verne, for example, correctly described in the late 1800s several of the technologies that now form part of our everyday life, including elevators, video-calling, and electrically-powered calculators that can communicate with each other over vast distances through messages, much like the internet.
Communication technology has evolved rapidly since then, with 5G cellular networks now in place across the UK. The next generation, 6G, and beyond are expected to rely heavily on ‘distributed networks’, a type of communication network that will enable many of our current sci-fi fantasy technologies to become a reality. So, what is a distributed network?
Our typical telecommunication networks take on a centralised structure, as shown in the diagram above. This means that any exchange between participants in the network travels via a central hub. For example, if Person A were to send a text to Person B, that text would travel from A via the communication link to a cellular base station, and then via another link to B. Similarly, payments usually travel through banks, and energy through the National Grid. The hub regulates the network, acting as a traffic controller and screening any exchanges happening between other devices in the network, or nodes, for security. However, the activity in a centralised network is limited by what the hub can handle, sometimes this limitation leads to bottlenecks in dense, busy networks such as in phone lines at midnight on New Year’s Eve. The hub also forms a major point-of-failure in the system and is therefore a security risk.
This point-of-failure vulnerability is evident in James Cameron’s Terminator film franchise: Skynet, an artificially intelligent, self-aware neural network, attempts to launch a nuclear attack against humankind. The “good guys” saved the world by attacking Skynet’s hub, i.e., the central device that defends Skynet’s other nodes. When the hub goes down in a centralized network, there are no communication links or operational protocols in place between the remaining nodes that allow them to continue communicating with each other, so the remaining network typically fails. In the third Terminator film, however, Skynet returns as a distributed network comprising of many interconnected computers across the world. It cannot be disabled in the same way as before because it no longer has a single point-of-failure, and to disable it would require taking down all of its infected devices. Thus, Skynet wins.
Mass human extinction aside, this contrast between the first and third movies highlights that decentralising a network can result in a more robust system. As the nodes work together in a cooperative and self-regulated way, there is no reliance on a single central authority and the network becomes more resistant to attacks.
Terminator also demonstrates how decentralising a network allows it to scale more easily. Skynet was able to spread its reach across the globe because each infected computer only needed to be connected to one other infected node in order to be a part of the network, rather than all needing to be connected to (within communicable reach of) the same central hub. This way, Skynet could span a much larger area than its centralised equivalent, a trait that will be very useful for large-scale applications such as smart city infrastructure, as discussed below.
Even in the real world, attacks on network hubs happen frequently. For example, in April 2021, a member of an online forum acquired and published the personal details of over 530 million Facebook users by hacking Facebook’s central database. Although the internet seems at first to be already distributed, given that everyone participates in it independently, the handling of information on the internet is highly centralised. Richard Hendricks, the fictional protagonist in HBO’s series Silicon Valley, attempts to tackle this issue by re-designing the internet from scratch. “We could build a completely decentralised version of our current internet,” says Hendricks, “with no firewalls, no tolls, no government regulation, no spying. Information would be totally free in every sense of the word.” Though idealistic, the mission of decentralising the storage and exchange of our personal information is plausible given that a similar leap has been made in the financial world: according to Fortunly, an estimated average of over £120 billion per day is exchanged via distributed cryptocurrency networks, the once-unthinkable alternative to our centralised banking systems.
Another useful type of distributed network is a wireless sensor network (WSN): a collection of many sensor nodes that cooperatively collect sensor readings in real-time, allowing for automated data acquisition over large-scale or remote areas. For example, in the 1996 natural disaster film Twister, a team of scientists invent a device that releases thousands of tiny flying sensors into the heart of a tornado in a mission to collect atmospheric readings from the vortex’s centre. These interconnected sensors work together to capture a bigger, more detailed picture of what happens inside a tornado than a single large sensor alone could have achieved.
The idea for Twister’s flying sensors came from a real-world project by the NOAA’s National Severe Storms Laboratory. Sadly, NOAA’s WSN, nicknamed TOTO, was never successfully deployed into a tornado and was retired before going on to inspiring the movie. There have since been numerous, more successful attempts and, in addition to monitoring extreme weather, WSNs are expected to become an integral feature of our biggest industries. One such industry is e-agriculture, where a WSN can be distributed across vast farmland to gather real-time information about the crops and to trigger a response mechanism, such as adjusting crops’ irrigation in response to fluctuations in humidity and soil moisture.
No futuristic landscape is complete without a ‘smart city’ – a metropolis of highly automated infrastructure including flying taxis and delivery drones zipping through the air. Futurama’s introduction sequence imagines exactly this, but the idea goes back decades, with Isaac Asimov’s 1953 Sally featuring a distributed network of cars and autonomous public transport. The core idea is that the city forms an ecosystem of different interconnected devices such as the vehicles, buildings, speed cameras etc. that communicate directly with each other and also make decisions without the need for human input. These automated, ecosystem-like networks are called the Internet of Things, or IoT. For example, the aim of vehicular IoT is to enable vehicles and infrastructure to make highly informed decisions autonomously on our behalf, such as vehicles collectively and safely negotiating the right-of-way between themselves at a busy junction.
There are, however, some technical challenges to overcome before distributed networks can be fully integrated into our everyday systems. One hurdle is the delegation of responsibility in the system – without a central authority, who allocates system resources? Who makes the final decisions? Researchers are attempting to develop the fairest, most efficient consensus protocols. These are rules to dictate how network nodes will make group decisions, usually optimised to balance all nodes’ best interests as well as the network’s functionality as a whole.
Another challenge is finding a reliable and practical way for the devices to independently assess each other for security purposes. Without a central authority to regulate the network, what happens when a node goes rogue or is compromised? Some distributed networks include powerful devices that can secure the network computationally such as the super-powerful computers that validate transactions in a typical blockchain network. However, the majority of devices such as phones, sensors, and personal computers don’t have the complexity to handle these computationally-intensive security protocols.
In Charlie Brooker’s Black Mirror, the episode Nosedive explores a world in which people rate each other on a scale of one to five after every interaction, such as after a conversation or based on appearance. A person’s overall rating then determines their leverage and opportunities in society, with higher-rated individuals having access to better services. This reputation concept can be applied directly to devices in a distributed communications network, and there is a whole field of scientific research devoted to developing secure trust inference algorithms, where the devices rate each other after interactions.
Trust and consensus are just two of the key challenges being tackled in the field of distributed communication networks but, already, one can see that these issues are almost human in nature. Perhaps, they signal the first steps towards the age-old fantasy where machines and humans are indistinguishably similar. We can already hold intelligent conversations with an AI algorithm via ChatGPT, and generate astonishingly detailed artwork using MidJourney. With the advent of machine learning, and as the technology for IoT nodes evolves, one can hopefully expect to see a resemblance between our next generation of communication networks and some of the more down-to-earth science fiction ideas within the next decade or two. For flying cars – maybe a little longer.