Decentralizing Cellular Networks

We’ve written several posts about the possibility of “re-decentralizing” the Internet (here and here) and this topic continues to attract attention, particularly with the hype (and corresponding backlash) around “Web3”. In this week’s post, co-authored by Amar Padmanabhan, we’re turning our attention to the possible decentralization of the “mobile packet core”, which promises a significant departure from how cellular networks have traditionally been implemented.

---

While the Internet’s architecture is highly decentralized—routers, switches, and autonomous systems are distributed across regions and providers—the cellular network has historically been run in a more centralized fashion. The typical mobile packet core comprises a small number of devices that are centrally located. The core is connected to a set of widely dispersed base stations using backhaul links.

This model is a good fit for traditional telcos, who obtain licenses for spectrum over wide areas and can operate services using a mostly centralized packet core that is appropriately sized for the customer base. But with the rollout of 5G, the wireless landscape is changing. Increasing bandwidth and density requirements are driving the adoption of “small cells,” which make use of spectrum with a range of about 10 meters to a few kilometers. This means that the set of radios connected to the cellular network will be both more numerous and more heterogeneous.

Small cells may also mean that many of these radios are no longer under the direct control of the telco. Small cell radios are typically much smaller than the giant cell towers used for long-range coverage. The smallest ones can be about the size of a home Internet router. Given appropriate access to radio spectrum, individuals or businesses can put up their own small cell radio base stations on the real estate they already own. This contrasts with the typical capital-intensive approach of large cells, in which telcos must acquire rights to locations on which to construct towers. In this way, small cells effectively drive decentralization.

In the United States, the availability of CBRS, a “lightly licensed” spectrum band, addresses the issue of access to spectrum. Now, with small cell radios and appropriate spectrum, we have two of the critical components for a decentralized cellular network. But how do we make it all work as a coherent network service? We need a wireless network architecture that can leverage a set of decentralized base stations.

The Magma project (as we have discussed previously) offers an approach to building mobile packet cores that is more distributed that traditional 3GPP approaches. Magma pushes much of the mobile core implementation out closer to the edge, with small access gateways sitting close to radios. Unlike traditional mobile solutions, Magma readily scales up in small increments with the addition of these small access gateways. Magma also supports federation with existing cellular networks, making it possible to authenticate users against different customer databases. With this model, multiple carriers can tap into this decentralized network to extend their reach without having to make large capital investments.

An important question remains, however. How do we incentivize people to deploy these small cells? For this we turn to the Helium Network, a novel approach to decentralized wireless networks.

The Helium Network

If we want decentralized networks, we will likely need a decentralized way to pay for their usage. The Helium Network leverages a “proof of coverage” technique to create tokens on a blockchain. Individuals and businesses that stand up a small cell receive Helium Network tokens (HNT). (As with most of the blockchain world, the value of those tokens fluctuates considerably.) HNT can be used both to pay for service and to compensate the operators of base stations without the need for a central entity (like a telco) to manage subscribers. Proof of coverage compensates those who stand up small cells in advance of the arrival of traffic, incentivizing the build-out of the network.

Although most Helium Network hotspots currently serve up LoRaWAN (an IoT-focussed wireless technology), Helium has recently launched a new 5G service, which leverages the Magma platform. Dish is the first major carrier to announce it will allow its customers to “roam” onto the Helium 5G network wherever there is coverage. The Helium Network uses the federation capabilities of Magma to access Dish’s subscriber database. If a user is authenticated by the Dish home subscriber server (HSS) and Dish agrees to cover the costs (using Helium’s native data credit mechanism), then the Helium Network will provide service to that user. Dish owns the billing relationship with the roaming user and ensures that the Helium network receives payment for the costs incurred. It’s worth noting that some entity will need to operate a federation gateway between Helium and Dish, but this too can be decentralized. In other words, there can be more than one such gateway operated by different entities.

It’s intriguing to see how a fairly traditional, centralized provider such as Dish can leverage a grassroots-operated, decentralized model like the Helium Network to extend service to its customers. Since the Helium Network is open and permissionless, there’s considerable opportunity for commercial entities to provide services directly to their users via payments based on the Helium blockchain.

It seems likely that the future wireless network will be a combination of centralized, service provider-led networks and more decentralized private- and community-deployed networks. To realize this future state, three pillars need to be in place: a cost-effective and decentralized mobile core; access to spectrum; and a decentralized economic platform. Magma, CBRS (or other lightly licensed spectrum), and the Helium Network, respectively, represent at least a first attempt at providing those pillars. It is time for service providers to think about how they can benefit from expanding their business models.

---

If you’d like to learn more about Magma, we’ve put together a course on EdX, and there is documentation available.