While never a major problem for cellular services so far, reliance on fixed infrastructure right up the radio tower has always been a slight constraint. It is harder for networks to cater for variations in demand over time without the ability to follow where the users are, which could vary from areas of dense population in the evening to busy business districts during the day. In the 5G era the static nature of the architecture could become more of a problem as demand grows for denser cells serving smaller numbers of people.
This is the thinking behind a proposal from a Japanese research group at Tokyo University of Agriculture and Technology to recruit private cars to house small radio units via crowdsourcing, offering some incentive in return. That incentive could be financial, based on a slice of revenues generated, or some other reward.
The selection of private vehicles was based on the idea that these were likely to be closest to where their owners were as a representative subset of the mobile subscribing population. This could only be roughly correct since many people use other forms of transport both for work and leisure, so the theory needed to be tested.
However, a study indicated that a hybrid network comprising fixed towers augmented by small radio units on cars could ensure reliable service with required coverage and capacity at a substantial saving in total hardware compared with fixed infrastructure alone. The idea could be implemented using vehicles owned and moved by mobile operators themselves, with the potential then to optimize locations, but that would impose a massive capex far greater than just deploying small cells in the normal way, say on street furniture.
It looks far more promising to exploit part of the consumer base of vehicles already on the road when no significant additional upfront costs would be involved. That could build on work already done for military and public safety applications, in which moving cellular base stations have been developed, housed in vehicles or even backpacks, and communicating with one another, platoon-style, when out of reach of the main RAN or core.
The Tokyo system envisages a far more mainstream and consumer-focused application. It was built using C-RAN architecture, with wireless fronthaul to connect the radio units on the vehicles to the shared baseband. As in most 5G C-RAN designs these days, the base station functions are implemented separately in three elements of the network – the central unit (CU), distributed unit (DU), and radio unit (RU), the last of these housing the antenna, power, analog/digital converter and other functions.
The DU, in this design, performs PHY and MAC layer functions, while the CU handles higher layer functions. Increasingly RUs are being deployed densely as small cells and that was the arrangement for this Japanese trial.
The experiment was designed for a smart city where roadside units (RSUs) would have been already deployed quite widely, to support communication with crowdsourced RUs (CRUs). Then edge servers control the states of the CRUs via the RSUs. The idea is that stable wireless fronthaul links are easily established with static channels by focusing only on vehicles that are parked. This sort of architecture would be needed in any case to support autonomous driving via vehicle to infrastructure (V2I) communications and on that basis the deployment would be relatively inexpensive and consume little additional bandwidth, while not being critical and therefore time-sensitive.
However, the network has to be operational with a combination of parked and moving vehicles. For this it used the already established multi-gigabit vehicular communication system originally developed for mobile connections to passengers in moving cars or trains, optimized for efficient allocation of resources in the presence of interference that varies rapidly in time in millimeter wave bands.
The CRUs themselves are then activated only when the vehicle is stationary, to save power as is important since they are charged by the car battery. When a CRU is available, a wireless mobile fronthaul link is established with a nearby distribution unit ready for transmission of data to phones in the vicinity.
The simulations were promising, comparing the proposed system with a traditional fleet of stationary RUs. It was found that 100 RUs on parked cars complementing 200 stationary units, making 300 RUs in total, could deliver better data throughput than 400 uniformly dispersed stationary RUs.
The Tokyo group believes this sets the stage for commercialization of the technique, although it will have to wait for the required roadside infrastructure to be available. It is certainly an interesting addition to the expanding panoply of small cell housing.