Extremely Low-Latency, Bidirectional, Multichannel Audio Design Using Software-Defined Radio (SDR)

As a networked, professional audio product design company, Cardinal Peak is dedicated to helping brands bring their new audio products to market quickly. For networked, multichannel, bidirectional pro audio solutions, latency is a major concern. Real-time applications, such as conferencing, live performance and interactive gaming, require low latency and consistent jitter so the audio feels natural to listeners and doesn’t cause cognitive jamming.

Low-Latency Wireless Audio Design

Achieving low latency in wireless systems is challenging since commercial wireless standards have significant and nondeterministic latency. From the table below, Bluetooth and 4G are not acceptable for low-latency audio designs. While there is potential for 5G and Wi-Fi, those systems introduce variable delays as networks become loaded. This variability is often a bigger problem than the absolute latency.

Typical Latency (ms) for Commercial Wireless Interface Standards

Software-Defined Radio Design Example

We had a customer that needed an extremely low-latency (submillisecond) link for both wired and wireless 10-channel bidirectional audio. For the wireless solution, we utilized our software-defined radio (SDR) audio expertise to develop a custom SDR with sub-ms latency using Analog Devices’ (ADI) Catalina transceiver. For the wired version, we used ADI’s A2B. The A2B wired and SDR wireless links were just one part of the overall networked audio system, which also leveraged Audinate’s Dante protocol for time synchronization as shown in the following figure.

We elected to use ADI’s Catalina transceivers for a couple of reasons. First was the ready availability of an RF SOM with a sufficiently sized FPGA to implement the radio along with an embedded ARM core to run Linux (Zynq). Second was the frequency agility, as we needed both 2.4 GHz and 5.8 GHz to accommodate data since our requirement was to be able to run multiple Catalina-based systems simultaneously. This frequency agility precluded existing solutions from companies such as Lime.

The challenges included the following:

• Implementing the radio entirely in logic (FPGA) instead of software to meet latency requirements (see the following figure)
• Designing a custom waveform
• Could not use Wi-Fi or other commercial protocols because of latency issues
• As we had a custom waveform, we could not leverage all the commercially available test tools for commercial protocols

In our implementation, all the modem processing was done in the FPGA instead of the microprocessors to improve throughput and lower latency as shown in the following figure.

While this project was implemented using Catalina, we see how the newer Navassa parts, with their lower bandwidth and lower power, will be well suited for pro audio wireless microphones. The lower bandwidth means that no FPGA will be required, so the radio could be implemented using a low-cost, low-power microprocessor. These same qualities also mean that Navassa is well suited for land mobile radio applications. Given Navassa’s new chip-to-chip phase synchronization capability, we also see some great applications in phased array applications, such as satellite tracking and radar.

ADI & Professional Audio Design Experience

• The solution described above leveraged our experience with Audinate’s Dante.
• Being an ADI Alliance Design partner, we have deep experience with ADI’s audio products, including SHARC and A2B as well as the SDR components like Catalina and Navassa.
• For more information on our A2B experience, along with descriptions of those projects, please see our A2B case study.