Note from the Editors: The following is the second in a series of guest articles by experts from across the broadband ecosystem. Bill Rose is President of WJR Consulting Inc., providing product and business development services to the home networking industry. A frequent speaker at industry events, Bill also chairs the Consumer Electronics Association’s Home Networking Committee and the Technology and Standards Council, is a board member of the Home Networking and IT (HNIT) Division, and sits on several corporate Technology Advisory Boards.
Entertainment Networking Requirements
Home Networking has arrived! WiFi or IEEE 802.11b based wireless networks are being deployed at an accelerating pace in homes; experts say that shipments of WiFi devices have exceeded wired Ethernet at retail outlets. However, these networks are primarily being used to share files and Internet connections. For home networking to reach the mass market, most experts agree it must support entertainment applications – especially audio and video distribution throughout the home. So where are the entertainment networks?
To answer that, we have to start with the requirements for an entertainment network. It is generally accepted by the consumer electronics industry that to be successful, an entertainment solution must meet a number of criteria:
No-New Wires Solutions
While structured wiring with Ethernet is the home entertainment network solution of choice for new homes, the cost and other issues with installing new wires make it unattractive to most homeowners.
Instead, “No New Wires” has become the holy grail of networking. HomePNA™ (telephone wiring) and HomePlug™ (powerline carrier) both claim to handle entertainment and they are correct – to a point. However, HomePNA has been around for a few years and has not made any market penetration so we can assume the market has spoken.
HomePlug is a more recent entrant to the market. With a realized throughput of around 4-5 Mbps in a home, it is an excellent solution for both data and audio networking but is too slow for high quality video.
Both HomePNA and HomePlug have announced plans for higher-speed solutions. Only time will tell whether they will be competitive with wireless solutions for whole-home entertainment networking.
The Wi-Fi family of wireless solutions consists of two physical layer standards – 802.11b (operating in the 2.4 GHz band), and 802.11a (operating in the 5GHz band), providing maximum data rates of 11 Mbps and 54 Mbps respectively. A new standard called 802.11g is in process, which will provide 54 Mbps in the 2.4 GHz band. We can quickly dismiss 802.11b with real throughput of 3-5 Mbps as too slow for HD and SD video. This leaves 802.11a and g, both providing a realistic throughput of up to 30 Mbps.
There are many good arguments that neither of these can handle video across an entire home based on their throughput and range. 802.11g networks, operating at 2.4 GHz, suffer from a congested and noisy frequency band populated by cordless phones, microwaves, audio and video repeaters, and other devices. Current 802.11a implementations suffer from higher signal attenuation and therefore less range; studies have shown that it can drop well below 20 Mbps with less than 50 feet and a single wall between devices. However, let’s assume for now that both can deliver adequate throughput throughout a home. The real problem is their dependence on CSMA/CA.
Networks and the data they carry can generally be placed into two categories: asynchronous and isochronous. The file you send over the network to be printed, or the Web page you browse is asynchronous data: a delay of a second or two is no problem. Live or real-time video, audio and telephony data is called isochronous: these applications only work properly when guaranteed that data will arrive on time.
All 802.11 standards use CSMA/CA (Carrier Sense-Multiple Access) at the Media Access Control (MAC) layer and are often referred to as “wireless Ethernet” because they are based on the Ethernet protocol, which is also CSMA based. CSMA requires a device to wait for the network to be idle (Carrier Sense) before transmitting. CA or Collision Avoidance was added to the 802.11 MAC to compensate for the fact that it is impossible to detect a collision on a wireless system. CA mandates that after detecting an idle line a device must wait a random amount of time before transmitting. Since both of these wait periods are variables, the result is an asynchronous network solution.
A highway provides a good analogy to CSMA/CA. The highway itself is the wireless channel the devices are operating on, and the cars are the data packets devices are trying to send. A car entering the highway has to wait for a gap in the traffic before it can get on. This is the carrier sense part of the protocol. Additionally, some highways have traffic lights at the entrance ramp. This further delays the cars entering while they wait for a green light. This equates to collision avoidance.
Carrying the analogy further, once you get onto the highway, there are no guarantees as to when you will arrive at your destination. A 65 MPH speed limit does not guarantee you will be able to go 65 MPH.
All 802.11 networks using a CSMA/CA MAC share these same characteristics. As the traffic gets heavier, it takes longer and longer to send data. The result can be a blank screen or dropped frames on the TV set. Adding a large amount of memory to buffer the signal can help but it costs money. It is also hard to determine how big a buffer is needed. If the network is too crowded, it continues to fall further behind. Additionally, it takes time to fill a buffer. If you like to channel surf, the buffer has to fill up each time you switch to a new channel resulting in a delay before seeing what is on. A 3 second wait results in 90 seconds to surf 30 channels.
To try to fix this situation, IEEE created Task Group “e” (TGe) several years ago to develop a “quality of service” (QoS) mechanism for 802.11. While continuing to use CSMA/CA as the basis for the solution, they added priorities (Enhanced Distributed Coordinator Function or EDCF) and an optional Hybrid Coordinator Function (HCF). EDCF is analogous to emergency vehicles: everyone else (lower priority data) has to wait while they pass. HCF is analogous to a traffic cop; it allows for more tightly controlled access to the network for high priority data, at the expense of all other data transfers.
While EDCF and HCF help, there is still no guaranteed arrival time possible. They must be backward compatible so there are still variable wait periods, and simply increase the probability of data arriving on time.
There is a better alternative. With TDMA or Time Division-Multiple Access at the MAC layer, each device with data or content to send is guaranteed the bandwidth it requires. QoS can be guaranteed because the network controller, typically the access point, grants a portion of the bandwidth to requesting devices. When there is no more bandwidth to give out, it will not grant the device permission to communicate until some is freed up. The network can also reserve some bandwidth for asynchronous data to ensure your Internet surfing experience does not slow down to that of a dial-up connection. Thus both isochronous and asynchronous data is carried simultaneously.
To accomplish this, TDMA networks divide the total available time into time slots and assign those time slots to a given application – say watching a movie. There is no need to wait for an idle line; a device just waits for its time slot and transmits.
TDMA is analogous to a rail system. You contract for a certain amount of space on a train (the time slot) and your cargo (the data) is guaranteed that space. Since the conductor (the access point) knows the train’s departure time, and the speed it travels at, he can guarantee the time of delivery.
Several wireless technologies are based on TDMA, including IEEE 802.15, HiperLAN2 in Europe, and AIR5™ from Magis Networks, Inc. IEEE 802.15.3, one version of 802.15, is being developed specifically to handle entertainment in homes and meets all of the requirements except for range and availability: it is only designed for a range of up to 10 meters, is not yet an approved standard, and no devices are available yet.
HiperLAN2 was specifically designed to handle entertainment. It supports all the requirements for entertainment networking including high throughput, support of both asynchronous and isochronous data, 3DES security, low error rates using forward error correction (FEC), QoS, etc. Because its genesis was from the access side, it supports many of the access side protocols such as ATM making it a costly (from a processing and memory perspective) standard to implement. At this time there are no chip sets available, nor does it appear to have any traction in North America, or for that matter, in Europe.
Recognizing HiperLAN2’s strengths as well as its shortcomings, Magis designed AIR5 based on a subset of HiperLAN2’s MAC layer -- making it much smaller, simpler and less expensive to implement. AIR5’s physical layer and RF design are based on 802.11a, but it includes many enhancements providing greater range and an improved ability to penetrate walls and other structures.
At the MAC layer, it supports all of the features in HiperLAN2 that made it excel at entertainment networking, without the additional access side features, which are not required for home networking. It also includes support for 1394 and 5C copy protection, something Ethernet-based networks do not.
In Europe, there is a requirement to support power control and frequency agility. Power control allows a device to use only as much transmit power as is needed to reach the destination device. Frequency agility allows it to automatically change channels to avoid other networks and noise sources. Together they provide for more efficient use of the available RF spectrum, and are included in both Hiperlan2 and AIR5. The IEEE 802.11h standard specifies these features but they are not included in currently available Wi-Fi implementations.
The bottom line is that wireless entertainment networking has arrived – but it’s not the Wi-Fi Highway. It’s the TDMA high-speed rail system.