Flexilink: A Low Latency Solution for Packet Based Media

Precise live audio collaboration over networks for professional applications has become practically (and theoretically) impossible with the widespread adoption of IP-based packet technologies. Traditional approaches used solutions based on the layer 2 (link layer). These protocols included AES50 [1] and MADI, and use a time-slot-based approach (similar to Time Division Multiplexing). However, the IP-based “best-effort” approach is subject to quality of service (QoS) availabilities in the network since data packets are required to queue in buffers in switching nodes as they progress through the network. This builds up an unpredictable degree of end-to-end latency for audio data through the network. This paper sets out a unified low latency network architecture that simultaneously supports both time deterministic media data and also the remaining best effort traffic; both traffic types can co-exist and be transported in the same bit-stream utilising a hybrid of “reserved-slot” (c.f. TDM) and best effort (unreserved) resources in the switch nodes. For the live production and presentation of audio, the proposed technology gives low latency performance as well as the support of multiplexed multiple channels each of which use different sampling rates and word lengths. The technology is also being considered for compressed media, including video.