• The operating environment,
• The functionalities,
• The physical issues, and
• The technical specifications.

Except for the storage capacity, the DPVR could be built today (but probably would not address a large market).

perating environment

Image quality
I expect that HDTV will be generally available a couple years before the holiday season of 2009,  at least for the high-end consumers using the Dream PVR. This high resolution is the level of image quality expected, even though, when probable improvements in image compression are taken into account, the overall amount of data to be handled per minute of video may not change that much compared to today's standards.Data feed

The dream PVR should be able to receive all sorts of inputs, beyond the usual cable and radio frequencies. In particular it should be able to operate successfully in a home environment, with at least the following types of inputs:

• Cable, satellite, radio,
• LAN (get data off the internet and the home PC /data center, including Gbit Ethernet and wireless Ethernet)
• High-speed serial connection, be it USB-3 or 1394b, which will allow a good connection to video cameras and transportable storage.

The various corresponding formats of data will be translated on the fly to a common internal format. This will eventually require that the opposite translation be done when playing the recorded content from various media.

• directly to a TV-like unit, with a specialized cable, or wireless connection
• using the home LAN
• using a high-speed serial connection

Some outputs may be specialized with a lower bandwidth. Examples include sound systems, and surveillance cameras displays.

In a home one could expect an entry-level model to support three users, while a high-level model would support six users or more. A hotel or a condo building could go to hundreds of users. This is beyond the scope of this paper.

Functionalities of the Dream PVR
Functionalities describe the features people would like to have.  These need to be simple and obvious enough that everybody will remember how to use them easily and efficiently. They include the usual record and play, but also some selection mechanisms for special features (such as "parental control"), as well as the allocation of storage priority -as the storage capacity albeit large is not infinite- this could include priority settings between new recordings and discards.  There should also be a capability to play items from locations external to the PVR itself, and also copying them to such locations, while keeping a directory of these items.  Following are a list of desired functional characteristics for the dream PVR.

Recording
The dream PVR must be able to record streams and stills, from a variety of sources (see "feeds" above), simultaneously, and while playing some streams (those being recorded or others). It should be able to record at least one stream per user at any time, preferably two streams per user. This recording is done on its internal storage space. Pushing items outside of the PVR is done in "copy" mode, as this implies gaining sufficient control of the storage space elsewhere and would be possible with an integrated storage network.

• The user defines if the recorded item is private or to be shared, and if so, whom with.
• When the storage fills up, the unit can either decide on its own what to discard, or ask for guidance from the appropriate user(s).
• This is a PVR, there is no major editing capability, besides deletion and splicing.

• already stored items (e.g. to decide whether to keep them or not), and
• on-going recordings (e.g. to decide to cancel the recording).

The administrator has access to all recordings, while an individual end-user only has access to the recordings that are his/hers (end-user deletion or cancellation of recording is not available for shared items).

Any user can select relative priorities for their items, as some of these will need to be discarded to give room to new ones.

Part of a recorded item can be spliced onto another recorded item, provided that both are under the full responsibility of the user.

• which feed to use for a new recording,
• which types of contents to choose for recording, with priorities, or
• which output to use for playing a recording (if not the one the user is currently using)

Passwords and quotas can also be set and changed by the dream PVR administrator or set by defaults.

• The eventual fan must be a very low noise fan
• The disk drive must be very silent.

The unit cannot generate a large amount of heat. One simple way to limit heat (and to reduce the need for a fan) is to limit the overall power consumption, probably to significantly less than 150W. (a typical cupboard lamp is 60W, and this can already generate a considerable amount of heat).

The unit must survive realistic conditions that occur in an unoccupied house, as well as operate in conditions that occur in an occupied house:

• non-operating: -5°C to +50°C, 10% to 99% humidity
• operating: +10°C to +50°C, 30% to 99% humidity

Dust can be an issue, the unit must be able to do its job in a dusty environment without any issue, be it electrical shorts or heating up.

• 8KV electric shock (kid dragging nylon socks over thick rug before getting a shock when getting close to the unit) operating
• one meter drop over rug, non operating

I would expect each user to want to store / keep about one year's worth of movies so that they are easily accessible, be it on the unit itself or catalogued by it (and then on some external storage). Make it 350 stored items per user, assume that 2/3 of those are shared, this yields

• for the low-end unit (3 users): (350 * 2/3 + 3 * 350 * 1/3)= 570 say 600 movies or shows,
• for the high-end unit (6 users): (350 * 2/3 + 6 * 350 * 1/3) = 920, say 900 movies or shows.

Other items, such as still pictures or sound tracks take very little space relative to movies, so are ignored at this point.

Reliability of the storage is typically linked to redundancy. Most drives operate four to five years without major issues, yet if the units are to reach the level of reliability seen in earlier CE video storage systems (such as DVDs or tapes), reliability may well become an issue. A "dream" PVR will guarantee (albeit with some possible interventions) that the contents are kept intact if the unit is properly maintained.

Items stored externally, say on a computer, must be readily usable by the unit, hence the unit can play, and trick play streams in various formats.

• low-end: 3,600GB, and
• high-end: 5,400 GB,

for a safe average of 6 GB per movie.

There is a caveat at this point: the above figures assume, as stated at the beginning of the article, that the compression algorithms will improve enough in the near future to almost compensate for the increase in image size due to the move to High Definition (HD) from the current Standard Definition (HD). If this is not the case, the above figures could easily be multiplied by two. It may be worth also at this point to consider the expected storage capacity of Hard Disk Drives in the coming years, as shown in the figure below.

Typical high-capacity drives should have reached 1.5TB by late '09.

• The super cheap models will offer one drive only (1TB to 1.5TB, no redundancy), this will badly need expansion capabilities.
• The low-end systems will probably offer two drives (for 3 TB, no redundancy), a little short.
• Middle of the line systems will offer four drives (6TB without redundancy, or 4.5TB with redundancy).
• High-end systems will offer six to eight drives (6TB and more with adequate redundancy).

Despite the expected progress in flash-based storage, there is little chance that their price point will be competitive with disk drives in those capacities at that point in time.

Given the perceived never-ending "need" for space, an important aspect of the file management system will be the capability to organize and find content, especially content that is not stored on the PVR itself, but stored somewhere on the local network.

Hard disk drives are quite sensitive to various kinds of shock and other environmental hazards, and their increased capacity will keep them this way.

HDDs in the dream PVR will probably have to be (physical and electrical) shock mounted, and have very good heat dissipation capabilities, so they can operate effectively in the high range of required temperatures and humidity.

A web-based interface seems natural for the unit administrator, and in the absence of a LAN, a simple screen interface should be made available locally.

A simpler interface is to be made available to each user, based on their local viewing medium.

• cable, satellite dish, regular TV and radio (AM and FM, satellite)
• LAN (this covers direct access to the web as well as wireless LANs),
• USB, 1394

The unit must be able to play or connect to many different units:

• TV, HDTV
• mp3 players (load'm)
• mpeg4 players
• LAN
• USB, 1394
• Media servers

If the mainstay of the connection is the house LAN, this will probably have to be upgraded to Gb Ethernet so it can handle the various feeds at once.

At least some of the mp3, mpeg4 and TV outlets will require the unit to have the corresponding encoding capability, and to do it on the fly.

A low-end unit will have fewer connections and capabilities than a high-end one, and will require less storage.  A lower end system will also not be able to store much HD content.

A simple estimate is to BOM cost a given connection at about $10, and a disk drive at about $150, then to add the casing ($20), the CPU ($20), and the board (another $20).  The units with enhanced reliability will also include some extra drive handling capability ($20)

The following table gives a rough estimate of the BOM and retail costs of the various units:

 
It takes about 18 months from inception to delivery for such a system. These will be available for the holyday season '09 if the corresponding programs are launched by early '08.

The major remaining issue is how to index and find content.  Solving this problem will be a boon to PVR-using mankind and will make someone rich!

# # # #