In early November, the ON-SIDE team gathered at the BBC Studios at Pacific Quay in Glasgow to test and measure the performance of various 5G architectures and video encoding.
As part of the project, we are investigating the use of 5G networks in a broadcast studio environment. While the BBC has deployed 5G networks to support production in the past, notably at the King’s Coronation in 2023, these deployments have focused on a contribution use case where multiple user devices are connected to a network as independent devices. As there is no interdependency between these user devices and other equipment, they have different requirements around latency, and they often use heavy compression to move audio and video signals over the network.
In a production studio environment, we have different requirements: we have more stringent latency targets, and the video quality needs to be as high as possible. This is in order to be able to inter-cut the output of a radio camera with other sources such as cabled cameras.
There are other considerations too: we need to be able to control a camera and communicate with the operator, as well as carry signals to and from ancillaries such as autocue or microphones.
Currently, these cameras are usually configured using dedicated radio links for each camera on a unidirectional path with dedicated spectrum for each unit. Add on all of the control and communication requirements, which also have dedicated radio resources, and we have a complex ecosystem that requires specialist equipment and engineering.
As part of the ON-SIDE project, we wanted to see if we could use private 5G networks to support wireless camera operations in a studio environment. This has the advantage of a single IP- based bidirectional configuration which means we can be more efficient in the use of spectrum. For a given spectrum channel of sufficient bandwidth, we can, in theory, operate multiple cameras and associated equipment.
In order to determine if 5G private networks could meet our requirements, we needed to look at several different aspects of the workflow. To address the latency issues, there are two main considerations:
- the time taken to encode and compress and then decompress a video signal;
- the time taken to carry this data over the network reliably.
For the contribution use case, where latency is not as strict, we can take time to compress the video and send it over the network and allow for any packet loss or retransmission of data. Typically, this is between 1 and 2 seconds.
In the studio environment, we need to aim for a latency of less than 100 ms. This is to enable not only inter-cutting of content but also a near-instantaneous response to any control signals to the camera. The latter is particularly challenging as we need to factor in round-trip time, not just the glass-to-glass latency.
For the testing, we set up three independent private networks: a Cisco core and an ORAN solution running in the N77 spectrum band and a Neutral Wireless Lomond NIB running in the N40 spectrum band.
Spectrum was a mix of ON-SIDE and BBC spectrum secured via Ofcom’s current spectrum licensing systems, and one of the aims of the project is to identify how we can streamline this process.
We also ran various encoders from several manufacturers at different bit rates and recorded the output so that we can do some later analysis of the encoder performance.
We measured the radio performance at dedicated points around the studio for each system and collected data for each of the networks.
We also configured camera control and other devices, such as mobile phones, to act as monitors or comms units.
The full results are currently being analysed and will form part of the output documentation of the ON-SIDE project.
The project team would like to express their thanks to BBC Scotland, Sony, and Haivision for the loan of the studio, encoders and cameras that enabled this testing.