Articles and whitepapers

Digital Streaming over IP - Moving Audio and Video Distribution Forward (2/2/2005)

By Alan Ainslie, Invision UK

Distribution and control of multiroom audio, with video as an AV experience, is at the centre of the drive towards smart homes, being the most clearly beneficial and attractive part of the offer for the average home owner. Conventional audio distribution by matrix switch however, with distribution at line level for room amplifiers, or distribution at speaker level with long speaker leads, and control by data cable, raises a number of issues.

Firstly, long speaker leads, or line-level leads, compromise sound quality. Secondly, an inflexible wiring infrastructure tightly defines the system's functionality, and any system changes require major wiring changes. Feedback to the user is also seriously limited and requires separate distribution for real feedback from sources such as hard disk servers. Finally, integration with other systems, such as lighting, requires either a proprietary master control system or specific interfaces and software.

A PC-based home network on the other hand, employs a well-understood set of conventions that requires no client specification. Digital data transfer guarantees data integrity. The network should be able to support all required client functionality, and many events can be transported simultaneously without affecting other networked experiences. Furthermore, an unlimited number of devices can be attached, and changing needs can easily be accommodated.

Residential IP networks

Home networks, often referred to as a part of a structured wiring environment, are now a standard part of the specification of a home. Driven mainly by the need for multiple access to the Internet, the wired infrastructure offers much more, including distribution and control of audio with full user feedback, and the ability to integrate with other similarly-connected systems.

Strictly speaking, we are talking of a proper meshed network, referred to as Ethernet. Its cabling would be CAT5, supporting a data rate of up to 1000Mb/s. The cable radiates in a star formation from a central switch, and the characteristics of the switch will have a large bearing on the network performance. Devices can either send data to, or receive data from the switch on a single CAT5 cable, and Internet access is simply a router connected to the switch on one side, and to the outside world on the other.

The universal protocol for all activity on Ethernet is TCP/IP (Transfer Control Protocol/Internet Protocol). Each device on the network has a unique IP address, and this can be allocated automatically and dynamically (DHCP) or can be manually allocated.

Network operation

Generally, data on a network comes in bursts and is exchanged between devices as packets, assembled into tidy 'fields' with destination and source addresses. The amount of information in a field is flexible, and IP allows any number of addressed devices to throw data onto a network to any number of recipients, hoping that the network is clear. In a normal network, there is the possibility of data collisions - the more data, the greater the probability. The data is therefore resent in hope of a clear slot in the network, and arrives complete within a field, but with an indeterminate delay before all fields that comprise the complete data have arrived. Transmission is 'broadcast,' meaning that data is sent everywhere, not just to a specific recipient.

Media streaming in particular, has specific requirements. It is continuous, dense, and has few addresses which change only occasionally, and is generally not required to arrive at ALL addresses. The data is still two-way, but there is the opportunity to add full user feedback - perhaps even in the form of Flash movies, although this will add to data flow.

A switched Ethernet network, with an IGMP (Internet Group Management Protocol) Layer 3 switch, can be used to give streaming media priority and direct it towards only the elected recipients, i.e. to multicast rather than broadcast, and management software can be used to keep all in order, manage any number of sources or zones on the network, and maintain quality of service.

Typical IP audio distribution system

A typical IP audio distribution system would be based around an IGMP Layer 3 switch which is used as the centre of a conventional CAT5E network. The devices attached to the network would include streaming sources, reproduction devices, user interfaces, and analogues sources.

The Netstreams SwitchLinx IGMP Layer 3 switch

Streaming sources

Streaming sources deliver data streams to the network. A single source may launch several simultaneous streams for different users. An example would be an IP-enabled hard-drive audio server or possibly a PC. Along with the data, the source should also deliver user information as metadata comprising play lists and cover art for example. The source will also need to respond to commands from any user interfaces to enable music selection, pause, skip etc. Each stream of the source will be assigned a unique reference or name.

Reproduction devices

In each zone we need to recover the streaming audio via an IP-enabled room amplifier with an IP address unique to that zone. The room amplifier recovers the data, performs D-to-A conversion, and drives speakers via an in-built amplifier. Additionally, the reproduction device will serve full status information, such as volume, tone, etc, for the user interfaces.

User interfaces

To control both source and reproduction devices, a user interface is needed. Again, this will be an IP device on the network, and will display the metadata from the streaming source as well as the parameters of the reproduction device. IP addresses are linked to allow control of specific zones from specific keypads, possibly by user menu.

Analogue sources

To launch an analogue source onto the network, an IP-enabled media input device will perform A-to-D conversion and packetisation, and will launch the data from that IP address. Control of the source requires that IR learning be accommodated, and the media input device needs to serve the correct set of buttons for control to be displayed on the user interface. The media input device could also serve status information.

The Netstreams MediaLinx media input device

Performance capabilities of the residential network

Although streaming MP3 data is much less demanding than bit-for-bit digital audio, with the latest technology there really is no need to compress the data, and so sound quality is not compromised. With proper management, it is quite possible for a residential network to support many tens of 1.4Mb/s streaming audio sources, from up to 256 locations, and fed to 256 zones.

The number of media streams from any one source is governed by its access time and processing speed, but currently 6 audio streams per source is easily achieved. For streaming video, gigabit networks are the order of the day, as opposed to 10/100Mb/s audio networks. Streaming video is accepted as being compressed, although we are currently looking at 10 - 20 high-resolution 1080i streams from any of 256 source locations to 256 display locations.

Hard-drive audio servers

While there has been a rapid growth in availability of IP devices, such as security cameras, desk and pocket PCs, and various control devices, the first really stable and easy-to-install audio distribution solutions have only recently come forward - driven mainly by the need to fully utilise hard disk audio or media servers.

Hard-drive audio servers pose an interesting challenge. Such massive storage requires the user to have easy access to the metadata which delivers the selection menus, play lists and cover art. While this can be achieved using the video output of most current servers, if 2, 4 or 6 simultaneous listening experiences are to be served, delivering the correct metadata to each zone by video creates all sorts of problems. These are currently best addressed by hierarchical control systems of the server and video matrixes by companies such as Crestron or AMX.

Hard-drive audio servers delivering multiple analogue outputs for each user also tend to get very expensive, requiring additional analogue outputs for each user, at additional cost. The obvious solution to these challenges has to be distribution of the media, metadata, and control over the residential Ethernet network using TCP/IP.

An example of a digital audio over IP system is DigiLinX. This was developed by Netstreams in cooperation with AudioRequest, and is distributed by Invision UK. DigiLinX is a suite of devices that operates on a residential network built around a two-way IGMP Layer 3 100Mb/s switch. The same network can be used for all Internet and file sharing activities in the home.

A standard AudioReQuest hard-drive audio server connects to the network using its Ethernet connection. The AudioReQuest can be configured to serve between 1 and 6 simultaneous full-bandwidth audio streams plus the corresponding 6 sets of metadata onto the network without any cost penalties - six users are no more expensive than one.

Audio is derived from the network by an IP-addressed room amplifier that is little larger than a packet of cigarettes. Being a TCP/IP environment, control within a room can be by a colour LCD touch pad, a pocket PC or a wireless web tablet using an access point connected to the switch. Local inputs can be enjoyed locally or from any room via the network. Sources requiring full IR control sit on the network courtesy of a media input device which not only performs A to D conversion and IP packetisation, but has a full specification IR learning engine that is ideal for SKY boxes.

The Netstreams SpeakerLinX IP-addressed room amplifier

Simple and flexible installation

All IP devices on the network require a connection back to the central switch. A zone may contain two or three IP devices, such as a touchscreen, an amplifier, and a media input box which are accommodated via 4-port switch in every touchscreen. This means that in most cases, a single CAT5 from the switch to each zone will take care of the data infrastructure, and will also provide PC access in the zone for Internet access.

While IEEE 802.3 allows for power distribution over Ethernet (POE), the maximum 15W is not really sufficient for any serious audio applications, and so we distribute 28V as well. To make this simple, we recommend that a shotgun CAT5E plus 4-core speaker wire type cable is used from the switch into all zones, and a power supply be located alongside the switch. Devices are located wherever the client requires, are easily moved, and can be added with a simple software update. Commissioning is a simple PC-based wizard-driven process that takes no more than 30 minutes to configure a large system.

Conclusion

Residential networks are the obvious way to distribute and control the AV experience - increasing both performance, flexibility and usability for the client. Ethernet offers a mature, stable, and standardised environment for delivering distribution and control solutions, and the freedom afforded by IP, as well as the benefits of a truly universal protocol, mean that the number of residential and commercial IP products will escalate, ensuring freedom from obsolescence for those forward-thinking enough to be installing IP solutions today.

Alan Ainslie is the Sales Manager for Invision UK, trade-only distributor of affordable, innovative, next-generation products and solutions for audio/video entertainment and home automation.

www.invisionuk.com