Design Article
A better way to implement video over WLAN
Gil Epshtein, Senior Product Manager, Metalink
11/2/2005 2:32 PM EST
A wireless networking technology is required that is capable of providing sufficiently high bit rates to support the distribution of multiple quality video and HDTV streams from a central location, along with total home coverage. Video applications cannot tolerate bandwidth fluctuations, so guaranteed bandwidth and Quality of Service (QoS) are also essential requirements.
Also, a wireless network should provide wire-like performance regardless of changing environmental conditions. These and other challenges in the new home environment cannot be met with existing IEEE 802.11 a/b/g-based wireless products.
Solutions that attempt to use WLAN technology for video distribution have fallen well short of consumer expectations for link range and picture quality. New techniques have been developed aimed at supporting high-quality video applications with higher bandwidth, lower jitter, reduced latency and extended coverage.
Higher throughput and extended reach
Although higher effective throughput is not the remedy for all issues related to video delivery over WLANs, it is a major step toward a robust solution.
The ability to deliver much higher throughput provides emerging wireless video solutions with much better immunity to interference, and the means to handle degraded link conditions. Excess bandwidth can also be traded for extended reach and lower power consumption.
Figure 1 illustrates a relationship between range and effective bandwidth. IEEE 802.11a achieves 25 Mbps effective throughput at less than 30 feet, which barely covers the living room of a typical home. Metalink’s WLANPlus, on the other hand, reaches more the 100 feet at the same rate, using the same 5 GHz band, which enables full coverage.
A series of techniques described below is used to achieve this.
MIMO technology provides breakthrough performance
MIMO (Multiple Input Multiple Output) technology is a signal processing and smart antenna technique for transmitting multiple data streams through multiple antennas and achieving higher rate, extended range, and better spectral efficiency than possible with legacy wireless systems.
MIMO is a multi-dimensional technology. It sends an independent data stream through each antenna, increasing the wireless spectrum utilization by a factor equivalent to the number of transmit streams (also known as the MIMO Rank).
To accomplish this, MIMO utilizes spatial multiplexing (multiple antennas) on top of orthogonal frequency division multiplexing (OFDM). Coding the information across both the spatial and spectral domains by using multiple transmit and receive antennas, combined with OFDM modulation on each antenna, increases the diversity and with it the robustness. This enables MIMO to withstand channel impairments such as inter-symbol interference (ISI) and other interferences.
Multi-path environment
On the transmission side, MIMO encodes a single high-rate data stream by splitting it and transmitting it across spatially separated antennas. Having two streams instead of one enables either the delivery of twice the throughput by keeping the same rate for each of the streams, or extending the reach of the original stream since each of the lower-rate streams can use lower constellations and require a lower signal-to-noise ratio (SNR) to be recovered.
On the receive side, the MIMO receiver uses algorithms to recover the transmitted signals and combine them into a single stream.
The main advantages of MIMO include higher data transmission rate by a factor equal to the number of transmit streams, and the ability to establish a wireless connection in multi-path environments.
Channel Bonding
Shannon's capacity law indicates that the theoretical capacity limit increases linearly with bandwidth. Therefore, the simplest solution to increasing the rate of any given system is to expand its operating bandwidth.
For wireless LAN systems, this is usually called channel bonding because the extended bandwidth is achieved by bonding two adjacent 20 MHz channels into a single 40 MHz channel. The bandwidth increase is actually more than double since the guard band between the two bonded channels can also be removed.
Figure 2 illustrates the results of channel bonding.
5 GHz band essential for improved performance
Two spectrum bands are available for WLAN communication: the 2.4 GHz band (2.4 GHz to 2.4835 GHz) used today by the IEEE 802.11b and IEEE 802.11g standards, and the 5 GHz band (5.15 GHz to 5.85 GHz), currently used by the IEEE 802.11a standard.
The 2.4 GHz band can accommodate up to three non-overlapping 20 MHz channels, which imposes hard limitations on the number of users that can be served and the number of adjacent networks that can operate without interference.
Since only three channels are available in the 2.4GHz band, channel bonding is not a feasible option. Adding to the challenge is the interference resulting from home microwaves, Bluetooth devices and cordless phones, which all operate at the same band. Given these channel-bonding and interference issues, the 2.4 GHz becomes irrelevant for quality home video distribution.
The 5 GHz band offers more than twenty 20MHz channels in most parts of the world, which allows the support of much higher numbers of users, much higher bandwidth per user, and higher immunity from interference.
Advanced Forward Error Correction Scheme
Metalink’s WLANPlus uses an optional low density parity check (LDPC) instead of the legacy convolution code used by other IEEE 802.11 standards.
An LDPC code is a linear block code specified by a very sparse parity-check matrix. LDPC provides coding gain which is higher by approximately 3 dB than the convolution code, and has already been verified and adopted by DVB-S2 satellite broadcast and 10 Gigabit Ethernet over Copper system specifications.
The additional coding gain can be used to extend the reach for the same data rate. For example, 3 dB of LDPC coding gain translates into up to 30 percent improvement in range. It can also be used to increase the throughput (by using higher constellation) or to increase the robustness and immunity to interference.
The coding benefit of LDPC is highest when low packet error rate (PER) and high data rates are required. This is exemplified in demanding applications such as video distribution.
The simulation results shown in Figure 3 are for a 2 x 3 MIMO system with channel bonding, using the ETSI channel A model, with output power of 13 dBm per antenna and SNR difference of 3 dB between LDPC and non-LDPC modes of operation.
As was stated before, the coding gain merits of LDPC enable higher throughput at a lower SNR. This SNR gain translates to reduced RF costs, increased rate or increased reach as described above.
Improved MAC Efficiency
The 802.11 MAC/PHY has fixed overhead irrespective of packet size. Reducing the overhead is one of the main concerns in enhancing the current 802.11 WLAN standards.
MAC efficiency of IEEE 802.11a/b/g is typically about 50 percent at the best conditions. MAC efficiency can be increased to 70 percent by using an aggregation scheme for packets assigned to the same destination, thus eliminating the overhead linked to each packet and replacing it with a common overhead.
Aggregate exchange sequences are made possible with a protocol that acknowledges aggregated MPDUs (A-MPDU) with a single block acknowledgement (Block ACK) instead of multiple ACK signals. This protocol effectively eliminates the need to initiate a new transfer for every MPDU.
The common overhead associated with each Multiple MAC protocol data unit (MPDU) transmission of IEEE 802.11a/b/g, is now associated with a large number of MPDUs. This proportionally increases the efficient throughput. Up to 32 MPDUs with the same destination address and priority are aggregated into a single concatenated payload, called an aggregated MPDU (A-MPDU).
Figure 4 demonstrates the MAC efficiency difference between a system that doesn’t use aggregation (MPDU per aggregate =1) and a system that does. The graph is shows the effective throughput versus the number of MPDU per aggregate for a PHY rate of 216 Mbps and with 1,000-byte packet length. It is clearly seen that a system that doesn’t use aggregation is blocked at 30 Mbps effective throughput (MPDU per aggregate is equal to 1).
A WLANPlus system using a maximum PHY rate of 216 Mbps, a 1,000-byte packet length and 32-MPDU aggregation can reach an effective throughput of 178 Mbps without bit error rate (BER) and 164 Mbps with BER of 1e10-5
Jitter Cancellation and Clock Recovery
Most video broadcasts use the MPEG2 transport standard. In order to be able to interpret the MPEG presentation and timing information in the right way, the decoder-clock needs to be locked on the encoder-clock. Otherwise, overflow and underflow might occur at the decoder buffer, which risks creating packet-loss.
Most common decoders cannot lock on the encoder clock when jitter exceeds 500 ns. When passing through asynchronous packet-network like a wireless network, a much higher jitter might be introduced and a jitter cancellation mechanism is required.
There are several mechanisms to cancel jitter and all share the requirement of requiring a common clock at the sender and receiver. Since the traffic is VBR, a jitter-buffer at the receiver-edge of the asynchronous network can be used in conjunction with timestamps signature at the sender-edge of the asynchronous network. These timestamp will inform the jitter-buffer mechanism exactly when to play-out the next packet.
However, interpreting the timestamps at the receiver exactly at the form originated by the sender requires clock-synchronization between the sender timestamp mechanism, and the receiver jitter-buffer timestamp mechanism. Otherwise, the jitter buffer can overflow or underflow and packets will be lost.
Many algorithms allow clock recovery between network ends in the presence of jitter. Since constant delay is not usually the case of asynchronous packet network, algorithms were developed to evaluate the network-jitter, and isolate it from the clock difference. Such isolation is required in order to be able to correct the sent timestamps so the difference between them reflects only the clock difference and not the network jitter.
Support of IEEE 802.11e QoS Standard
The existing 802.11 protocols primarily use the distributed coordination function (DCF) access method to the wireless medium. The DCF provides an equal chance to each device to access the wireless medium. When dealing with video, gaming and other applications that are intolerant to bandwidth fluctuations, the fairness access provided by DCF is inadequate.
The IEEE 802.11e standard is targeted at addressing these issues and contains two main sections. The first is enhanced distributed channel access (EDCA), which defines four priority levels or four access categories (ACs) for different types of packets. It doesn’t, however, guarantee bandwidth, jitter or latency.
The second is hybrid coordination function controlled channel access (HCCA), which guarantees reserved bandwidth for packets classified based on EDCA by using a central arbiter for the bandwidth usage.
While in the DCF all stations try to access the wireless medium with the same priority, in EDCA there are four levels of priority or ACs. The mechanism of listening to the medium and using a back-off mechanism to determine the allowed transmission time is similar to that defined by DCF.
However, unlike DCF, the maximum back-off times are different for the different ACs, meaning that higher-priority ACs have a shorter maximum back-off time than lower-priority ACs. The shorter maximum back-off time allows the higher-priority AC to win access to the wireless medium more frequently than the lower-priority AC.
Applications or packets that share the same AC also have the same maximum back-off time and, hence, the same chance to gain access to the wireless medium. EDCA is fairly simple to implement, but cannot guarantee latency, jitter or bandwidth.
A better solution
HCCA uses another approach to guarantee QoS. Instead of waiting for an idle time for transmission and using a back-off mechanism, HCCA relies on centralized control by the access point that can guarantee time and duration of transmission for each of the connected stations.
Every station that would like to join the network must request permission from the central access point. This request includes a traffic specification that details the QoS required by the station.
The access point then determines if it can support the requested QoS specifications and admits or denies a station. The access point maintains a centralized schedule that is based on the QoS requirements of all of its registered stations. Then, the access point notifies each of the stations about the time it will have access to the wireless medium.
Since this process is managed from a central location, it is guaranteed that the access will be contention-free.
Because everything is predetermined upon registration, HCCA is able to guarantee bandwidth, jitter and latency, which is otherwise a difficult challenge in a mixed data and multimedia environment.
Conclusion
The emerging IEEE 802.11n standard based on MIMO technology is aimed at meeting the challenges of distributing video in a home environment. MIMO technology is at the core of this next-generation approach.
The use of MIMO enables higher data transmission rates by a factor equal to the number of streams and the ability to establish a wireless connection with no line of sight. Better SNR compared to a legacy single input single output (SISO) systems can enable developing wireless video solutions to extend reach as compared to legacy approaches.
Enhanced performance and much better immunity to interferences is achieved through channel bonding, advanced error correction schemes in the form of LDPC, and the use of the 5 GHz band.
Additionally, features including new aggregation schemes to improve the MAC efficiency and support the latest QoS standards can also improve the overall performance. These features provide increased throughput, range and robustness in the face of interference, and create an enhanced, reliable user experience.
As discussed in this article, each of these features has its own merits. However, only the complete package combining all the necessary elements can provide the breakthrough abilities required to push the home broadband network to the next level, enabling reliable video delivery over WLAN.
Gil Epshtein is a senior product manager at Metalink. He is responsible for its WLANPlus product line. He holds a B.Sc. in Electronic Engineering from the Israel Institute of Technology (Technion) and has 15 years of experience in the telecommunication industry. Prior to joining Metalink in 2001, Gil worked for Teledata Communications and ECI Networks. He can be reached at GilE@metalinkBB.com.
