No comments

ABSTRACT

As the industry transitions from SDI based production to an all-IP studio environment progresses, some of the finer points related to a smooth migration such as Time & Sync are capturing more attention from the early adopters.

Phase & frequency alignment of baseband signals are a critical element in media production. In the IP world, the required functionally is delivered via the IEEE 1588 Precision Time Protocol (PTP) specification.

Whilst having enabled many industries to transfer their synchronisation requirements via PTP to the IP centric environment, special care needs to be taken for each specific industry and its specific constraints.

This paper draws on the extensive research the authors have carried out on the use of PTP for the media production industry. It summarises their work on areas such as how PTP aware vs. non-aware networks behave under load, IP Quality of Service for PTP messages and Grand Master redundancy models. Concluding with the impact of design con- siderations, network architecture constraints and device requirements for a successful all-IP synchronised media production facility.

INTRODUCTION

With the advent of distributed systems where every node has a sufficient amount of local resources to perform given tasks partly independently from all other units, reliable and timely data communication became a mandatory requirement together with the need for a common notion of time or at least a method to convey a common frequency to all nodes.

Lacking a unified communication mechanism fulfilling all demands, every application domain independently developed legacy systems tailored to their specific needs.In industrial automation, for example, a number of competing field bus systems were prevalent for more than 20 years.

In the broadcasting industry the great majority of studios still operate entirely with an SDI based infrastructure.

Common to all those communication systems was their ability to convey both data and some kind of time and/or frequency information to all nodes. These systems were highly optimized thus requiring dedicated hardware and firmware to be developed, maintained, and updated. This was also their most crucial shortcoming, as they could not keep pace with the ever growing demand for bandwidth combined with increasingly stringent requirements on low latency data transfers.

Whenever incremental improvements, by replacing only a number of core hardware units, were not a feasible solution, the entire infrastructure had to be replaced in a time consuming and costly process.

In recent years, Ethernet (and IP) based communication systems have significantly matured from their pure office IT based origins. Nowadays, they are being considered a cost effective yet highly powerful alternative to nearly any legacy system in the market, thus replacing these gradually for nearly any application domain.

Consequently, the broadcasting industry is moving towards the all-IP studio as well.

Leaving aside its many undisputed advantages over all legacy systems, Ethernet has one important property to consider: It’s an inherently asynchronous medium, to be more precise, data transfer is only synchronised on a per link basis between two adjacent nodes, thus Ethernet does not provide a common frequency on its own via the physical layer alone.

This turned out to be a shortcoming only at first sight. Legacy systems like SDI generally are limited to provide only a common frequency, as opposed to packet based communication mechanisms which can transport absolute time information highly accurately as well using dedicated protocols: NTP, the network Time Protocol and PTP, the Precision Time Protocol being the two most commonly used.

Version 2.0 of PTP – defined in the IEEE1588-2008 [1] standard – turned out to be ideally suited for highly accurate clock synchronisation over local area networks and even shows remarkable performance of wide area networks.This standard was written with a broad view for a variety of possible use cases without focusing on a specific application.

This was accomplished via a simple yet robust design while providing wide operating ranges for all relevant parameters to choose from. As PTP can easily and precisely be tailored to specific requirements via a PTP profile, it was rather quickly adopted by all relevant industries. Telecom [2, 3, 4], the power industry [5], and finally the broadcasting industry [6, 7] all defined their own PTP profiles and subsequently started to deploy PTP on a large scale, most of them as their only means of synchronisation.

The following sections will present a brief introduction of the basic principles of PTP complemented with the main sources of error to reckon with and efficient methods to cope with them. Special focus will be put on deployment strategies for large networks followed by a discussion of PTP redundancy.