Application of voice extension VoIP technology in Wireless LAN (1)

According to media reports from Hong Kong and Taiwan, with the improvement of the WiFi standard and the decreasing of the size of 802.11 chips, the function continues to expand, and the feasibility of the wireless Regional Network voice (VoWLAN) telephone system is also gradually improved. Dual-band mobile phones can use WLAN connections to provide reliable in-house voice services, while broadband telephone services connect notebooks through WLAN. On the other hand, the network phone number based on WLAN can easily support multiple mobile phones with only one WLAN base station, which is no inferior to the traditional wireless phone with a low-cost advantage.

The 802.11 standard establishes the basic mechanism required to provide a reliable and high-performance Wi-Fi network telephone system. The notable examples are security (802.11i/WPA) and QoS (802.11e/Wi-Fi multimedia ). In addition, single-key security settings such as JumpStart for Wireless of the Atheros open program code allow all users to display letters and numbers on their mobile phones, you can still quickly set the configuration of a WLAN phone.

One of the unstandardized projects in the WLAN network telephone system is the polling method ). Therefore, this paper discusses the advantages and disadvantages of the existing two round robin methods, and focuses on the power consumption, the most critical factor in mobile devices.

All methods to reduce power consumption must enable user devices to use low-power sleep modes as much as possible, the 802.11 chip must consume the lowest possible power in sleep mode to support this practice. The 802.11 chip must consume the lowest possible power in sleep mode. For example, the AR6000 mobile radio frequency single-chip (radio-on-a-chip mobile; ROCm) device of Atheros enables a sleep mode with extremely low energy consumption, and the Automatic Power-saving Delivery (APSD) technology. ROCm also provides excellent performance, enabling high-speed transmission to shorten the sending/receiving time, while embedded processors on the chip are self-contained drivers, distributes and processes regular network maintenance operations on the host processor. With the above practices and other power-saving strategies, the ROCm chip can improve the power consumption efficiency of WLAN operations, which is up to six times better than the traditional WLAN Chip, thus improving the battery life. A new generation of 802.11 devices that can implement various VoIP applications now include such chips.

Import voice to WLAN

802.11 WLAN can use high-performance components to provide reliable overall performance. However, this media feature is still facing severe challenges when processing voice traffic. Because WLAN uses a license-free spectrum, it must tolerate a large amount of interference from different external devices and other WLAN. In addition, WLAN does not support synchronous operation, just like other IP networks ). Therefore, it is usually impossible to make predictions in microseconds. Because VoIP is a fixed digital rate (CBR) application that generates VoIP packets (that is, frames) at a fixed interval, the CSMA competition law of WLAN obviously lacks the central synchronization sequence (centralized synchronous timing ).

This phenomenon makes a greater comparison with the Standard Telephone Mechanism Implemented by the mobile phone system. The mobile phone system uses licensed spectrum and well-planned base station deployment to minimize radio interference. The mobile phone system is synchronized from the phone to the backbone line, so it can grasp the time series at the microsecond level and never deviate from each other. Therefore, it can predict the size of the capacity, and the capacity is provided to a single service design application: speech.

The features of these mobile phone systems make it easy to comply with the G.114 standard recommended by the ITU-T, which specifies that the end-to-end latency budget cannot exceed 150 microseconds. Because the overall architecture of the mobile phone system applies time-pulse voice packets in a customizable manner, it is not necessary to ensure low latency, but to ensure that the voice packets are of special service quality (QoS) the Mechanism is prioritized. The mobile phone system uses the current slot, multi-job and voice service management to join the data service.

WLAN is just the opposite. voice services must use the functions originally designed for data. WLAN can only use a portion of the end point with a latency of 150 microseconds. If both ends use a WLAN for a conversation, the delay budget will be further reduced. In addition, if the voice packets must span the Internet or busy enterprise networks, the packets cannot be delayed or sometimes even arrive. Late packets may arrive in groups.

People who have used legacy Transcoder to communicate via voice over the Internet or universal WLAN will be familiar with these issues. One of the ways to establish high-quality VoWLAN is to change WLAN to meet the needs of traditional encoders. In fact, whether it is full-time or time-based, the exclusive implementation shows that 802.11 MAC can be changed to a synchronous, time-trough TDMA approach; this approach can effectively solve the problem of voice transmission through WLAN, however, such systems are usually incompatible with existing WiFi devices and networks.

Although the fully synchronized network is attractive, the lack of strict synchronization is also the main strength of 802.11. Over the years, we can see the advantages of this type of IP network in the competition between Ethernet and ATM networks. When reliable and adaptive (good enough) channel access is strictly ordered (perfect), satisfactory approaches are generally more popular due to diversity.

Another reason to avoid using synchronization when designing VoWLAN systems is that these systems are not operating in a closed environment. The main selling point of using WLAN to transmit voice is to enable dual-mode mobile phones and other voice devices to utilize the existing WLAN infrastructure.

Next-generation Decoder

One of the ways to improve the existing infrastructure of 802.11 is to use a newer voice decoder developed for Internet applications. These decoders greatly simplify VoWLAN design. The inefficient Internet telephone environment facilitates decoder development and achieves sound voice quality at an extremely low speed.

For example, the iLBC decoder, the core of the popular Skype network telephone system, provides features equivalent to the high-end ITU G.729 decoder. The ITU decoder only uses 8 kbps and provides voice quality like a public phone; the iLBC decoder from Global IP Sound requires a bit rate of slightly higher-13.3 kbps. Global IP Sound says their encoder speech quality is superior to that of the PSTN, and it can withstand up to 30% packet loss. Internet Engineering Research Team (Internet Engineering Task Force; IETF) has set standards for this decoder. CableLabs is used to specify the PacketCable Audio/Video Decoder specifications of the multimedia terminal adapter and media gate as necessary.

With this decoder, the necessary VoWLAN voice quality is easier to implement, and it can also solve the latency and Jitter Caused by the Internet. Therefore, it is particularly suitable for non-synchronous open systems such as 802.11. Since the decoder is so flexible, why should we develop complicated time series and Synchronization Methods?

Challenge Power Consumption

Although the decoder is so flexible today, timing sequence is still very important because it has a significant impact on power consumption. The synchronization feature of the mobile phone system enables it to easily and directly implement sleep/wake-up scheduling on the mobile phone. The mobile phone can safely enter sleep mode between packets. However, 802.11 of devices never know when to receive unexpected traffic, or have to respond to access points for other reasons.

Although there is a difference between a mobile phone and a VoWLAN system, the latter must make its battery life comparable to that of a mobile phone. Dual-mode mobile phones use the same battery for both types of functions, so they are bound to compare with each other.

Speaking of this, we can't help but want to synchronize WLAN operations. If the access point knows when the mobile phone enters the sleep mode and transmits the data only when it is ready, the mobile phone can change to the sleep mode on a regular basis. Access Points do not need to be transmitted to the mobile phone immediately when the VoIP boxes arrive. If necessary, these boxes can be placed in the buffer zone.

Currently, there are two operation modes, which can implement a good power-saving timing technology with enough synchronization in 802.11 WLAN, so you do not need to perform full synchronization operations. These modes include HCF Controlled Channel Access (HCCA) Controlled by the 'hybrid Control Function; HCF) 'and Enhanced Distributed Channel Access; EDCA ). Both models are part of the QoS specification in the IEEE 802.11e standard, and both of them can be used for power-saving communication methods in development, the data is transmitted at a fixed rate between the Access Point and the platform, without the need to synchronize the entire WLAN.

Synchronize with HCCA

The HCCA mode is like the N-body synchronization mechanism. The CBR round robin schedule is set for N platforms from the Access Point. Although the typical 802.11 system has no regularity, the platform should be synchronized by schedule as much as possible. It is reasonable to describe such a configuration as the N-body system, because the timing interference to the first station of the polling schedule will affect the timing of other N-1 stations.

When the AP receives CBR requirements from the platform through the traffic specification (TSPEC), The HCCA mechanism will play a role, and then the AP will communicate with the station for CBR scheduling. Once the AP accepts the Platform as a polling user, the platform usually goes to sleep until the AP's expected downstream polling or polling and VoIP boxes arrive (). Within the specified time period (the 802.11a/g architecture of OFDM is 9 μs, and 802.11b will be longer), the above line of VoIP data (or QoS-NULL) on the platform will respond. If the platform sends upstream data, the AP responds with ACK.

To know the power consumption efficiency of this mechanism, let's first consider the time ratio for the platform to maintain the wake-up status. If the HCCA mechanism needs to operate correctly, the platform must wake up from sleep mode before the AP performs downstream round-robin. Depending on the hardware design, it takes about 0.1 to 1.0 microseconds to wake up the program. Then the platform must wait until the downstream Round-Robin arrives, and the round-robin may not arrive at the expected arrival time of the platform. Different causes include interference, long-lasting frames on the channel, internal scheduling conflicts in the AP (polling other platforms), and higher-priority operations (the AP must transmit a Beacon), the first frame exceeds the expected switching time or the relative time shift between the AP and the platform, will cause delay. However, once the downstream Round Robin arrives, the scheduling will become predictable. Based on the selected decoder and PHY rate, the uplink/downlink message frame switching should take place within less than 1 microsecond.

In the HCCA mechanism, the uncertainty of the time sequence mainly comes from the delay of CBR polling schedule, the possible retry upon failure, and the change of transmission time when the variable PHY rate is used. Based on these uncertainties, the wake-up time for the platform is about 2 ~ 5 microseconds. In a decoder cycle of 20 microseconds, the wake-up sleep ratio is more than 75%.

HCCA fixed bit rate schedule

▲: The access station can implement the HCCA operation mode specified in the 802.11e standard, providing a predictable VoIP round robin schedule, so that the management can be performed when the WLAN platform can reduce power consumption in sleep mode.