In an effort to make a better wireless network, the Cambridge MA-based company BBN Technologies announced last week that it has built a mesh network that uses significantly less power than traditional wireless networks, such as cellular and Wi-Fi, while achieving comparable data-transfer rates.
The technology, which is being funded by the Defense Advanced Research Projects Agency (DARPA), was developed to create ad hoc communication and surveillance networks on battlefields. But aspects of it are applicable to emergency or remote cell-phone networks, and could potentially even help to extend the battery life of consumer wireless devices, says Jason Redi, a scientist at BBN.
Mesh networks – collections of wireless transmitters and receivers that send data hopping from one node to another, without the need of a centralized base station or tower – are most often found in research applications, in which scientists deploy hordes of sensors to monitor environments from volcanoes to rainforests. In this setting, mesh networks are ideal because they can be deployed without a large infrastructure. Because they lack the need for costly infrastructure, mesh networks can also be used for bringing communication to remote areas where there isn’t a reliable form of electricity. In addition, they can be established quickly, which is useful for building networks of phones or radios during a public emergency.
While mesh networks have quite a bit of flexibility in where they can be deployed and how quickly, so far they’ve been less than ideal for a number of applications due to their power requirements and relatively slow data-transfer rates. All radios in a mesh network need to carry an onboard battery, and in order to conserve battery power, most low-power mesh networks send and receive data slowly – at about tens of kilobits per second. “You get the low power,” says Redi, “but you also get poor performance.”
Especially in military surveillance, the data rates need to be much faster. If a soldier has set up a network of cameras, for example, he or she needs to react to the video as quickly as possible. So, to keep the power consumption to a minimum and increase data-transfer rates, the BBN team modified both the hardware and software of their prototype network. The result is a mesh network that can send megabits of data per second across a network (typical rates for Wi-Fi networks, and good enough to stream video), using one-hundredth the power of traditional networks.
To make the mesh network more efficient, the BBN team looked at three areas that could be improved. First, they modified the hardware in each node by building more energy-efficient radios. They split the radios into two parts: one uses little power and sends data short distances and slowly, the other uses more power but is able to blast data over kilometers and at a high rate. Most of the time, says Redi, the nodes communicate using the low-power radio, while the high-power radio sleeps. The high-power radio comes to life only when needed to send a lot of information or a powerful signal over a longer distance.
Second, the team looked at the algorithms used to synchronize communication between nodes. These algorithms are needed because, most of the time, radios in nodes aren’t sending or receiving data, just listening for it, an activity that “burns up power,” Redi says. The BBN researchers designed a protocol so that each node transmits data about its presence, and listens for other nodes, at particular times. In effect, the protocols synchronize the nodes’ watches.
The third trick that BBN employed involves using a different kind of algorithm. The researchers designed protocols that track the data traffic on the network and signal the nodes to modify their activity accordingly. This adaptability is key to saving power, says Abraham Matta, professor of computer science at Boston University, who’s working on the BBN project. “You have this spectrum of adaptability,” he says. “You adapt to different levels of activity,” from networks in which only a few nodes are sending and receiving small amounts of information, to networks in which almost all the nodes are passing around megabits of data. By instructing each node to collect and respond to the traffic and signal strengths, these protocols direct traffic along the most efficient path. For instance, depending on the traffic around a node, it might be more efficient for it to send out a quick, strong signal to reach another node far away, rather than to continuously send a weaker signal to a busy node nearby.
While none of these approaches is completely new, this is the first time a mesh network has used all three of them to produce such power savings, says Redi.
“It’s extremely impressive,” says Mani Srivastava, professor of electrical engineering at UCLA. He cautions, however, that the power efficiency could change as the network configurations change. But, while further work needs to be done, the power reduction that the BBN team reports is “a big deal,” he says.
Even so, BBN doesn’t yet have plans to commercialize the technology. Redi says that DARPA plans to field test mesh networks based on the technology next year.
Elements of the research could also be implemented in consumer devices. While mesh networks would not replace established cellular networks – the industry has already spent a significant amount of money on the infrastructure – some of the power-saving techniques could be modified to help make more-energy-efficient consumer mobile devices, says UCLA’s Srivastava.
For instance, elements of the split-radio technology used in the mesh network nodes could be applied to cell phones and could significantly increase standby time in phones or PDAs. And as phones increase in complexity and come with more and more functions, battery life is a weak link. “If you look at smart phones, the battery life is terrible,” Srivastava says. So modifying the hardware and software in a way that’s similar to the modifications in the low-power mesh network could make a difference. “Just imagine if power was reduced by a factor of 100 in commercial products,” he says.
DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.
“This is a profound moment in the history of technology,” says Mustafa Suleyman.
What to know about this autumn’s covid vaccines
New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.
Human-plus-AI solutions mitigate security threats
With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure
Next slide, please: A brief history of the corporate presentation
From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.