I remember how excited I was coming home from the first 1394 Developer’s Conference in the summer of 1997. It was a relatively close-knit group of only a few hundred attendees (and just a few vendors), most of whom had known each other for at least two or three years prior to that (at least through email) working on what we considered the future of home connectivity dubbed P1394. The “P” stood for Proposed, as in a specification that was not yet ratified by the Institute of Electrical and Electronics Engineers, Inc. (IEEE). The first actual release of the IEEE Standard for a High Performance Serial Bus (1394) was approved by the IEEE in 1995. 1394 finally was real, and I haven’t been the same since.
The primary theme of the conference was the promise of a “single connector” that would satisfy all of the interconnected audio, video, and networking needs of the home consumer into the foreseeable future. There weren’t many “real” products being shown at the time – mostly a few video cameras, some 1394 interface cards for the PC and Apple, lots of chip sets for engineers to use in developing products, and a variety of cables and connectors for hooking them all together. However, there was a vision that was crystal clear.
IEEE 1394 was to be become the preferred connection technology for handling multiple streams of audio and video content between all of the consumer electronics (CE) devices in the home. At the time, 1394-based technology would allow clusters of devices to connect together as little media or entertainment centers residing in various rooms in the home. It was based on 4.5 meter cables that would ship content around by daisy-chaining devices like TVs, set top boxes, video cameras, color photo printers, and PCs together.
These clusters then could be connected through the use of 1394-based hubs strung throughout the house. It even had the ability to power certain devices so it eliminated the need for “wall wart” transformers hanging off of them. The underlying protocols allowed all of these disparate devices to “see” each other and be able to talk together without the user being a rocket scientist or even a network manager or engineer. It was the first sign of true plug and play capabilities being brought into the reach of the average consumer.
Even then, there already was talk of the next generation of 1394 looming on the horizon that would allow those clusters to be connected together over longer distances at even higher speeds for delivering the ultimate dream of true distributed audio, video, communications, and control services anywhere within the house – just by plugging into this single 1394 connector.
IEEE 1394 has come a long way since that first developer’s conference. In fact, we have moved from the original 1394-1995 specifications to the more efficient and feature laden IEEE 1394a-2000 version. Most people now know 1394 commercially as Apple’s FireWire or Sony’s i.LINK technology. Almost all new PCs and Apples come standard with at least one of the full-fledged six-pin versions of the 1394 connector. There also is a “consumerized” little four-pin version incorporated into most of the newer digital video cameras on the market. It functions the same as the six-pin version, it just eliminates the extra space required for delivering power over the same 1394 cable.
There are several good references available on the web detailing the technical aspects of 1394 and how it works. I have included several of what I consider to be the better links at the end of this document. For most of this article, I will assume the reader has at least some familiarity with 1394 and technically how it functions.
What Makes 1394 So Special?
Essentially, 1394 was designed specifically to satisfy the connectivity requirements of consumer applications, entertainment devices, and content distribution services within the home. Since its inception, it has flourished to become the accepted digital interface standard for numerous other standards organizations including the Consumer Electronics Association’s (CEA) R4.1 and R7.4 Versatile Home Network (VHN) subcommittees, the Digital VCR Conference (DVC), the European Digital Video Broadcasters (DVB), and several others. For the consumer, it means getting rid of the tangle of assorted and confusing individual wires currently required for connecting many of the modern day entertainment devices together and moving to “that single connector” that just plugs in and “makes it all work.”
Dave Wooten, the (as he puts it) Proud Chair of the IEEE P1394b Working group, says that “I really think the requirements for connecting things together in the home are very different from the requirements in business. Basically, the digital home is not a small-scale version of an enterprise network. The use of the wires/fibers in the home primarily will be devoted to entertainment with some limited use for communications.
This is in direct opposition to what is going on in businesses, which are very concerned about their employees being able to communicate, but not so interested in the employees being entertained. This, it turns out, is not a trivial difference. 1394 is designed to allow arbitrary conglomerations of devices to connect together and organize themselves into something useful. To me, the really cool thing about 1394b is that when you conceive of such an application that requires this, you don’t have to throw your hands up in disgust because 1394b isn’t any help.
Rather, 1394b solves a large part of the problem. Sure, 1394b doesn’t cover every darn application instance, but it does provide a tremendous starting point from which an incredible number of application solutions can be developed.” I also thought I would mention that I can track emails back to early 1997 when Dave already was planting the initial seeds of 1394b to the IEEE. In other words, he has been living this technology for quite some time. It now has been over a decade since the original work on 1394 began at Apple.
There are a wide variety of technologies encompassed by the 1394 specifications. To me, there are five main attributes of 1394 that really help deliver on the vision of a true multi-purpose bus for the home:
Unsupervised cabling – 1394 allows the consumer to just plug devices in. The topology supports daisy-chaining from device to device, running the cables to 1394 hubs in a “home run” fashion, or a combination of both. The user does not have to worry about setting node addresses or providing name servers. It uses a peer-to-peer hierarchical memory/bus architecture and does not require a host for control. There also are specifications for using 1394 as a backplane architecture, although there currently are very few actually in production.
Plug and enjoy – Devices that plug into the 1394 bus are “discovered” automatically and their resources made available to all of the other devices. Most of this is through a related standard called the IEEE 1212 Control and Status Registers (CSRs). It is what allows the devices to communicate with each other through a standardized addressing scheme, control the aspects of each device using well-defined control and status registers, and interact via standardized transactions.
Isochronous data streams – I consider this to be one of the more important attributes of 1394 for the delivery of content throughout the home. It allows multiple devices to open “channels” across the wire with each having a minimum guaranteed latency. Although 1394 uses fairness intervals to ensure each connection stream gets the appropriate amount of time on the bus, isochronous data streams allow logical source and destination (sink) “plugs” to have a certain amount of time allocated to them so that audio and video can stream between the end points without something else hogging the wire (the stream won’t stop or stutter if a large data transfer is taking place at the same time).
1394 guarantees each talker a quantity of time and the talker has to allocate the amount of time it needs on the bus. If the talker does not use its time, or does not use all of it, the other isochronous talkers arbitrate for it. Any time not used by the isochronous talkers is given to the asynchronous devices so the time is not wasted. The interface that is used for setting up the channel “plugs” uses what commonly is called the IEC-61883 standard (I will mention more on this later).
Normal asynchronous “data” traffic also is supported, but it must “squeeze” in between the time actually used for the guaranteed low latency (isochronous) traffic. Isochronous traffic may use up to eighty percent of the available time on the bus. Asynchronous mode transfers focus more on the reliability of the data being transferred and support retries if appropriate.
Isochronous delivers low latency bandwidth and is more concerned about getting a particular type of media stream to a specific destination within an allocated time slot than it is on requiring retries and acknowledgements for guaranteed packet delivery. The 1394 architecture supports up to 64 isochronous data channels on a single bus, depending upon the bandwidth (the allocated time) required for each one. Note that even a single minimally compressed full motion video stream from a 1394 camera requires consistent bandwidth of about 25 Mbits/s.
A standard ATSC 720p or 1080i, 16:9, and 5.1 Dolby Digital stream from a D-VHS deck requires upwards of 19 Mbits/s through a 1394 interface. Keep in mind that we ultimately are talking about shipping several of these streams to different A/V devices located throughout the house over a single bus structure.
Standardized programming interfaces – Part of the 1394 specification defines how to “talk” to the various devices on the network through standardized programming interfaces. There technically are several interface layers that cover the physical and link level commands for controlling bus management and arbitration. Other programmatic interfaces are up at the application layer and are used for controlling and monitoring the A/V attributes of each device object.
Most programmers either deal with the inherent lower-level AV/C (Audio/Video and Control) interfaces or through one of the higher-level programming languages such as the one defined by the Home Audio/Video Interoperability (HAVi) group ( www.HAVi.org ). Although HAVi and 1394 sometimes are used in the same context, HAVi sits on top of 1394 and takes advantage of its services. It is not part of the actual 1394 specifications.
5C Copy protection – 1394 supports content copy protection at the packet level. This is of tremendous benefit to the consumer electronics industry. 5C copy protection (also known as DTCP or Digital Transmission Content Protection) provides for the encryption of content using key exchanges to ensure the viewer has the right to view and/or record the content coming across the wire. Some argue that copy protection infringes on their freedoms, but the entities who own the content want to ensure they also get their piece of the action.
They are supporting technologies like 1394 because it allows them to maintain control over their intellectual property. Because 1394 natively supports things like copy never, copy once, and copy always, it is a natural connectivity solution for devices like digital video recorders and media/content servers. It probably is worth mentioning that the newer content-encoded output switching feature provided by the D-Theater option for a D-VHS deck allows studios to encode the software for enabling or disabling the player’s outputs, which essentially allows them to control the random copying of tapes. Yet they allow playback through the 1394 port if it going straight to a TV set.
Major companies have made 1394 and its associated services the value-added selling feature on their new CE-based products. One of Mitsubishi’s biggest advertising themes lately for its WS-65869 65″ HDTV Projection Monitor is their NetCommand? user interface that allows one to watch and control any device connected to the 1394 network. They have included 1394-based software within the set to “automagically” generate icons on the TV screen through which a user can interact with a particular device using a standard remote control ( www.mitsubishi-tv.com/netcommand.html ).
The TV takes advantage of the application-level features of 1394 using HAVi, the device control and programming interface mentioned above, in addition to leveraging the standard AV/C capabilities inherent in 1394. These built-in technologies allow for the discovery of new 1394-based devices and their capabilities. The subsequent control of them is made available to their user interface through the standardized set of exposed object interfaces afforded by 1394.
NetCommand also takes advantage of 1394’s built-in 5C copy protection features (which is a much touted feature by them and the rest of the consumer electronics and content provider folks). When combined with other HAVi-based 1394 devices like their digital VCR (D-VHS recorder), Mitsubishi is one of the first major CE manufacturers helping to fulfill that dream of a single connector that allows one to connect and control all consumer electronic devices within the home.
As a side note, 1394 also supports a Serial Bus Protocol (SBP-2) for interfacing with hard drives. This allows one to connect a disk drive directly to a 1394 bus and have it available to any device that wants to use it for storage. Also, the current version of 1394 supports what ends up being about 400 Mbits/s. Not too shabby, even by today’s standards.
The Next Generation is on the Horizon
Even with all of the support for the current versions of 1394 by the consumer electronics manufacturers (and the entire electronics industry), 1394 has not been standing still. A lot of work has gone into making it even better. Last month the IEEE P1394b Working Group ratified the latest generation of 1394 technology dubbed 1394b (remember that the current rendition is 1394a-2000).
The Ballot Review Committee (BRC) for the Working Group finally gave its approval on this latest generation of 1394 (after a third round of final modifications and votes). This completed draft specification was approved (as I write this) by the IEEE Standards Association on March 21, 2002. Now that it is approved, it gets reformatted and goes off to publication. It will be available as an official document through the IEEE known as the IEEE 1394b Supplement to the IEEE 1394 Standard (with no P) – although I imagine it officially will be known as IEEE 1394b-2002. The P1394b (Proposed) draft document currently is available at www.zayante.com/p1394b/drafts/p1394b1-33.pdf. Reading it may not be for the faint of heart, but it does define every aspect of this newest release in complete detail.
IEEE 1394b promises to broaden the vision of delivering high-speed connectivity throughout the home by:
- Extending the allowed distances between device endpoints.
- Supporting a wider range of media types.
- Upping the throughput from the current 400 Mbits/s to what effectively becomes 800 Mbits/s, 1.6 Gbits/s, and beyond to 3.2 Gbits/s.
- Providing improved bus arbitration, power management, and speed negotiation.
- Facilitating lower RF emission designs.
- Opening up additional technology platforms for tackling some of the new and exciting markets like automobile “telematics” and airplane multi-media and content delivery networks.
All of these new features are made available while still maintaining backward compatibility with the plethora of 1394 devices currently on the market.
As mentioned earlier, the current versions of 1394 are limited to individual cable lengths of 4.5 meters. This is great if you want to connect a set top box to the TV right below it or to the digital VCR next to it, but it becomes a major limiting factor when trying to connect clusters of equipment together throughout the entire house. The 1394b version of the standard now supports the same multi-media connectivity over Plastic Optical Fiber (POF) at 200 Megabits/s (S200) and up to 50 meters, 100 Megabits/s (S100) on Unshielded Twisted Pair (CAT-5 UTP) to 100 meters, and up to 3.2 Gigabits/s (S3200) over Glass Optical Fiber (GOF) at 100 meters.
The improved speed and distance capabilities of 1394b result from two major improvements: overlapped arbitration and advanced data encoding. The new bus arbitration scheme (which determines what gets to have access to the bus and when), known as the Bus Owner Supervisor Selector (BOSS), implements what is termed overlapped pipelined arbitration. In other words, the protocols that manage and control the speed of the data as it traverses across the 1394 bus can run in parallel with actual data transmissions themselves. This is different than the way the 1394-1995 and 1394a legacy versions operate, both of which must alternate the information flow between data transmissions and arbitration protocols.
The new 1394b data encoding improvements are based on the same “8B10B” codes (a way to encode the data so that it is easier to send it over longer distances) used by Fibre Channel and Gigabit Ethernet. 1394b adds more robust control codes and actually scrambles both the data and control symbols resulting in much lower average Radio Frequency (RF) emissions. Together, the two improvements comprise what is called the Beta mode of operation (which is new and unique to 1394b) to distinguish it from the 1394-1995/1394a legacy mode’s Data Strobe (DS) coding (with the data traveling on one pair of wires within the 1394 cable and the synchronizing strobe on the other pair).
The recommended fiber to use with 1394b is 50 micron multimode Glass Optical Fiber. This differs from the 62.5 micron multimode fiber that most people have been installing for the 10/100 Mbits/s Ethernet world. 62.5 micron fiber also is what almost all of the current “Node Zero” structured cabling infrastructure vendors have been supporting.
When asked about the use of 50 micron multimode fiber instead of the more common 62.5 micron multi-mode, Dave Wooten says that he “Really does not expect anyone to start ripping 62.5 micron fibers out of walls so that they can install 50 micron fibers. I believe that the 50 micron will be necessary in order to get to S3200 (~3.2 Gbits/s), but I think that a lot of the installed fiber can be used quite comfortably at S800 and S1600. In fact, I hope someone develops a proposal for the signaling and connectors to allow 62.5 micron GOF to work. I also hope that someone will bring forward a proposal for use of graded-index POF for speeds above S100 and distances up to 100M.”
The good thing right now is that both 1394b and 1 Gigabit Ethernet technically can run over either 50 micron or 62.5 micron glass fiber. The distance for 1 Gigabit over 62.5 micron gets limited somewhat to about 300 meters instead of the 600 meters specified for 50 micron, but either should allow for long enough runs in most homes. What it really means is that, if 50 micron Glass Optical Fiber is installed in the house now, it will be able to support either 1394b or 1 Gigabit Ethernet. If you currently have the 62.5 fiber in place, it can be used for shorter distances with the right couplings attached.
What’s New About the Connectors Themselves?
With the availability of all of the new features provided by 1394b while still supporting the legacy versions of 1394, how does one distinguish which mode of 1394 will be used when you plug in a newer device? To solve the problem, the 1394b Working Group came up with a new connector for the native 1394b Beta mode. The physical layer interface of the new 1394b standard supports any combination of the following:
- Legacy ports – Those connectors which interoperate with devices built to the existing 1394-1995 or 1394a standards.
- 1394b Beta mode ports – The new connectors unique to 1394b which implement the higher speeds, longer distances, and overlapped arbitration.
- Bi-lingual ports – “In between” keyed connector plugs and sockets which can be connected to legacy, Beta mode, or bi-lingual ports on other devices.
The 1394b Proposed draft spec details the Beta mode connector’s various design aspects in excruciating detail. The “pure Beta” copper connector is almost as small as the existing 1394 4-pin connector, yet it is capable of carrying power, negotiating speed capabilities for a particular media type, and is capable of carrying a full 3.2 Gigabits/s signal to 4.5 meters. From a 1394b chip and hardware design perspective, the same chips can be used for all media types; it actually is just the front-end, connector, and analog components that are different.
It is important to note that the 1394b specifications do not encompass wireless connectivity. The 1394 Trade Association (TA) is working to make sure it is possible to encapsulate 1394 packets and send them over a wireless infrastructure. In addition, the 802.11 Task Group E recently adopted the QoS-enhancement proposals made by the 1394 Trade Association’s Wireless Working Group. This normally requires additions to Ethernet’s Quality of Service (QoS) features that most of the wireless LAN technologies support. HiperLAN2, primarily seen in Europe, does have a 1394 convergence layer defined already. Also, Magis Networks just started showing prototypes of their Air5? technology for 1394 over a very high-speed wireless connection.
As we migrate to 1394b devices that use the newer Beta mode only, we should see that they actually may be cheaper than the current 1394a devices. The Beta-only version, using what is termed unidirectional arbitration signaling instead of the legacy 1394 common-mode signaling, reduces the required voltages and allows for much simpler analog designs and reduced chip die sizes. This allows 1394b to use more advanced chip processing techniques, which ultimately require less total silicon to implement (although the actual gate counts for the 1394b chips will be about double the current ones – somewhere between 20,000 and 25,000). However, the 1394b encoding and decoding processes are more complicated for the chip makers to implement and require the logic and a Phase Locked Loop (PLL) to do the more complex 1394b clock and data recovery from the 8B/10B encoded stream (the legacy 1394a Data/Strobe encoding is simpler). Analog designs do not scale to the same degree as pure digital designs using a silicon process. In other words, the investment to implement 1394b into newer products probably will be about the same as for the current versions of 1394.
Where Will We See 1394b Used Initially and When Can We Expect To See It?
Max Bassler of Molex, Inc. the Vice Chair of the 1394 Trade Association says that “1394b, the newest version of the multimedia bus, is ready. It provides the bandwidth, distance, and overall performance to make 1394 the ideal choice for home networking, for many industrial applications, and for a new generation of computer peripherals. 1394b also is gaining momentum among the leading vehicle manufacturers who want to use it as the network bus for consumer electronics in automobiles. We expect to see the first 800 Megabits/second devices this summer and accelerated design activity by the autumn.”
“The advent of ‘b’ also will stimulate new applications for and development with IEEE 1394. These include the automotive sector, where the 1394 Trade Association’s Automotive Working Group has defined a widely accepted specification for in-vehicle consumer electronics networking. The in-vehicle video/audio network includes a customer convenience port that lets passengers connect their CD players, games, and other 1394-equipped devices and peripherals to the network with one plug-in cable.” The current implementation plan is to deliver these new 1394 capabilities over Plastic Optical Fiber running throughout the car.
Another key application enabled by the ‘b’ version is for “casual” storage, the kind of hard drives and mobile storage devices users hook up at random times. Most hard disks and optical media drives such as CDs, DVDs, and CD-RW drives require more than 5 Watts while spinning up and seeking. The 1394b standard provides significant amounts of bus power (unregulated 30 Volts DC at up to 1.5 Amps) along with aggressive power management so that power is used only when it is needed. Users with portable computers and peripherals that want to use casual storage devices will not need a battery-powered drive or some form of a “power brick.”
I asked Burke Henehan from Texas Instruments, a major manufacturer of 1394a chip sets and also a leader in 1394b chip production, how he sees the progression from the current 1394a products to 1394b. He replied that “The companies who already are doing 1394a probably will maintain 1394a for a while. Their moves into 1394b will be for the bi-lingual 1394b to ensure interoperability with their existing machines in the marketplace. Those starting new are looking at 1394b, with connectivity to legacy 1394a. It is the closed systems (automobiles, RAID boxes, etc.) that are thinking Beta-only. Even those vendors would like at least one bi-lingual node to communicate with existing products.
1394b will make inroads where it offers an advantage of throughput and/or distance or optical connectivity over 1394a. TI did a technology demonstration at CES 2002 in January showing a DV camcorder being controlled and playing video to an off the shelf Windows XP PC across a 1394b UTP-5 connection. There also are 1394a camcorders, Digital VHS, DTVs, STBs, and more being sold today with more planned for Christmas. 1394b 800 Mbits/s is being put on the market today. However, 3.2 Gigabits/sec is a formidable design challenge both from a silicon point of view and a board design point of view. It will not be completed until it is seen there is an excellent chance of getting a payoff for the investment.”
He also noted that “There are ‘structured wiring’ vendors who see 1394b as the overall interconnect for media centers and are developing products to go in the walls with 1394a being the interface coming out of the wall plate. This enables existing 1394 consumer electronics devices to be ‘home network ready.’ Then there are leading edge companies that are planning on putting 1394b in as the primary interconnect for all their devices – with TI doing both 1394a and 1394b, either is OK with us. For closed systems (cars, industrial systems, etc.) that are new, there is a bias toward 1394b (because of distance or optical interconnect) with at least one node of 1394b bi-lingual to connect to existing systems.”
And Now For A Little Personal Commentary
In addition to its native-mode communications protocols, 1394 does support running IP over it. This makes it an excellent networking platform. Several of the newer operating systems, especially from Microsoft, take advantage of the ability to run IP over top of 1394. In fact, the newer XP-based code loads an IP stack onto a 1394 interface by default during setup. This means that if you do not have Ethernet jacks on your computers, but they do have 1394 interfaces, you can connect your machines together using 1394 and take advantage of 1394b’s higher speeds and longer distances.
It easily could spell farewell to 100 Mbits/s Ethernet in the home. There also is a recently RFC created defining IPv6 over 1394. However, I personally believe there are some caveats with doing this from the 1394-based view of the world. The biggest problem is that IP is not the native-mode way for current A/V devices to communicate in a 1394 environment and most of the current 1394 products do not understand IP – which means they cannot communicate with IP-only based devices that may be plugged into the bus. They can share the same bits on the wire, but the upper levels will not “see” or understand each other.
Let me back up a little bit first. I originally started this essay with the intent of introducing the new features of 1394b and to help define some best practices for making 1394b the primary connectivity infrastructure within the home. This included how to wire for it, incorporate the appropriate fiber runs into a structured cabling design, allow for power distribution, and even how to support it from a development standpoint.
Since I have gotten into writing this, however, I have fallen into what has turned out to be a very heated debate with several of my cohorts over 1394b versus Gigabit Ethernet. I now am taking a little more cautious approach to implementing 1394 everywhere, especially as we look at the “recent” convergence of network-centric and PC-based technologies with the A/V and black box-centric solutions brought to the forefront by 1394.
I will preface my comments by saying I am comparing 1394b to switched Gigabit Ethernet, not the usual 10/100 Mbits/s Ethernet currently in use for most networks, especially in homes. With that said, the 1394 community is moving forward once again and is looking at using a 1000 Base-T PHY (Physical Layer) for the next revision. In other words, there already is work going on to allow 1394 to be able to support Gigabit CAT-5 UTP, the same as Gigabit Ethernet, while still providing all of the other benefits of the 1394b protocols. Fortunately, it appears the merging of the two worlds is beginning to shape up already, at least from a wiring infrastructure viewpoint. I believe it will be a while before the CE-based A/V products and the IP-only products can see each other, let alone be available in the marketplace.
The real issue centers around whether the requirement for isochronous data streams is real or is it just a perceived need that does not take into account the recent advancements in content delivery, compression, and lower prices for RAM-based buffering to avoid image jitter when using a strictly IP over Ethernet-based transport.
With the isochronous streams in 1394, the destination device acting as the sink for the stream plug does not perform local buffering and simply renders the data stream as it is received from the network. It can be argued that a buffered data stream loses the immediacy of a remote control action and that the user will not tolerate delays when performing transport control functions like starting or pausing a stream.
The techno geek in me says it seems reasonable that, if I want to ensure high-quality synchronized full motion, full screen video and surround sound audio with multiple streams being transferred simultaneously, leveraging the inherent isochronous capabilities built into 1394 makes sense. They are available now and have lots of support from the movers and shakers in the A/V marketplace.
The vendors taking advantage of them are the same ones manufacturing the products that deliver those media streams from the cameras, DVD players, set top boxes, tuners, TVs, and the like. On the other hand, there has been huge progress from the Internet-oriented and computer-centric factions of the industry when delivering quality video and audio from not only media servers in the home, but transparently receiving those same streams across the Internet over high-speed connections and even wirelessly.
To understand it from a very strategic view, I have been conversing with Bob Frankston (www.frankston.com), who is one of the most outspoken industry pundits pushing for an all IP world that I know (and admired for his convictions to not only move to IPv6 immediately, but to take the plunge and do it right at the onset by rolling out Encrypted IPv6). Anyhow, Bob says that isochronous data streams are not necessary for delivering high quality content. He sees 1394 as a closed platform that has not taken into account how the IP, Ethernet, and Internet industries have changed over the past few years and firmly believes that 1 Gigabit Ethernet will deliver as well if not better in the end. I, as well as most of my associates, also come out of the computer networking side of the industry and have IP protocols stamped on our tee shirts. So I can relate to what he is saying.