What is All-On-Board?
Whether you are building a PC-PBX, and need audio conferencing and a few channels of FAX, or are building a voice services platform and using G.168 compliant echo cancellation and full duplex audio streaming, you can do it all on a single card using iPline Technologies’ advanced All-On-Board technology. All-On-Board is about putting you, the application developer, in control. You control the combination of DSP applications and the number of channels served by each application. Any combination of DSP applications can run on a DSP. If you need more DSP processing power, you can add DSP modules without using any additional system slots. iPline Technologies delivers a wide variety of DSP applications and if we do not already have a DSP application that you need, we can develop it for you, or if you prefer, you can develop it with support from iPline Technologies.
A digital signal processor (DSP) is a type of microprocessor – one that is incredibly fast and powerful. A DSP is unique because it processes data in real time.
The core technology that makes all of this possible is our DSP operating system, OpenVPOS. OpenVPOS is a Co-operative Multitasking Real Time Operating System (RTOS) designed from the ground up to meet the demanding requirements of telephony signal processing.
OpenVPOS
OpenVPOS is an open RTOS designed specifically for telephony signal processing. OpenVPOS’s process scheduling algorithms allow OpenVPOS to support DSP applications with a wide variety of real-time response and scheduling requirements. Support of position independent code and data provides the flexibility to ensure DSP applications will never contend for the same physical memory, and ensures that a DSP can run any desired combination of applications. Support for a C-language programming environment makes it easy to port existing applications to run under OpenVPOS and simplifies DSP application development, whether we do it or you do it.
There are two classes of DSP telephony applications. One class is pass-through applications and the other is terminating applications. Pass-through DSP applications are those applications that form part of the audio signal path between two or more users. Pass-through applications include DSP applications such as audio conferencing and digital gain padding. Terminating applications terminate or generate a telephony signal. Terminating applications include audio play and record, DTMF detection and tone generation. Pass-through DSP applications have much more stringent real-time requirements than terminating applications. Therefore, pass-through DSP applications are often designed to run on a different DSP than terminating applications. For this reason voice card vendors will often require that different cards be used for audio conferencing and other DSP voice processing applications. The innovative process scheduling algorithms used by OpenVPOS allow both classes of DSP applications to exist on the same DSP. OpenVPOS supports pass-through applications with a signal latency as small as 0.375 ms. while running any combination of terminating applications on the same DSP.
Unlike most advanced general-purpose processors, DSPs typically do not provide hardware support for advanced memory management functions; for example, support for logical to physical address translation. Most DSP application developers respond to this limitation by developing DSP applications that use a predetermined block of physical DSP memory. This is fine if the required combination of DSP applications is known when the DSP code is developed and built. However, if DSP applications are developed independently, with the required combination of applications determined by the system application developer, then different DSP applications can end up requiring the same physical DSP memory. Because of this problem, many voice card vendors can provide only a limited number of DSP application configurations. OpenVPOS supports the development of position independent DSP code and data. Therefore, a iPline DSP application and its data can reside anywhere in physical memory. As long as there is enough physical memory, any combination of applications running on any number of channels can be supported. With our DSPs, you are not limited by the DSP application configurations.
Support for C-language programming, and the ability to create position independent code, allows iPline Technologies to provide an open DSP development environment. You can develop DSP applications that will work with our library of DSP applications, or if you prefer, we can develop the DSP application for you. iPline MonteCarlo 6.x contains a built-in API that allows you to communicate directly with any OpenVPOS DSP application. Therefore, any new DSP applications developed using OpenVPOS can be used with the current iPline voice card development environment, MonteCarlo 6.x.
Scalability
We use the Motorola DSP56303 on all of our products. This means that the same DSP applications can run on any member of our family of voice cards. Our default configuration provides one DSP for every T1 or E1 span on our digital cards, or one DSP for up to 24 channels on our analog cards, providing enough DSP processing power to meet the needs of most system applications. You can add additional DSPs by using iPline V-Engine modules (daughter boards) to provide as much DSP processing capacity as needed. You only pay for the DSP processing power that you need.
DSP Resource Calculator
We provide a DSP 56303 Performance and Resource calculator, which enables you to determine how much DSP power your system application requires. The resource calculator allows you to specify the combination of applications and channels that you will be using, along with those features that will have a significant impact on performance. The calculator determines the number of MIPs and the amount of DSP memory that will be needed. If there is enough DSP memory and enough DSP MIPs, you can be confident that your configuration will work. There are no other restrictions. If there is not enough DSP memory or MIPs, you simply add more DSPs.
DSP Applications
iPline Technologies provides a wide range of DSP applications. The philosophy of providing the maximum amount of flexibility and control is not limited to OpenVPOS but extends to each iPline DSP application.
Audio Play / Record
iPline audio play and record allows you to record using any of the 26 different audio formats. These include 3- and 4-bit ADPCM, Mu-law, A-law, 8-bit linear and 16-bit linear. Each of these encoding rules can be applied to data sampled at 4, 6, 8 or 11 KHz. Microsoft and ETSI 6.10 GSM are supported at a sampling rate of 8 KHz.
Other features include:
Automatic gain control (AGC) for both play and record. You have complete control over AGC performance. The gain adaptation rates, target output signal level, gain range and attenuation range can all be changed from their default settings.
Pitch corrected playback. Playback rates can be adjusted from 0.25 to 4 times the normal playback rate.
Voice activity detection (VAD) can be used to disable the recording of silence. You can control the VAD trigger levels and the debounce durations.
Pre-speech buffering can be used to ensure that audio immediately preceding the VAD trigger is not lost. You can control the amount of pre-speech that is buffered.
Play and record gains can be set to any value from –40 to +24 dB.
The size of the DSP audio buffers is completely under your control, as is the rate at which recorded data is passed between the DSP and the host. Data can be transferred in buffers ranging from a size as short as 6 ms up to a maximum size that is limited only by the amount of available DSP memory.
DTMF Detection
The iPline DTMF detector has excellent cut-through performance and scores 0 hits on the Mitel talk-off test tape. It has been thoroughly field tested to ensure the highest levels of performance. Though it is unlikely that you would need to, several DTMF detector parameters can be adjusted; for example, allowing you to change the trade-off between talk-off and cut-through performance.
Tone Generation
The iPline tone generation application allows you to define a nearly unlimited number of single and dual frequency tones. 60 tones can be globally defined and an additional 4 tones can be dynamically defined for each channel. You can specify the duration, amplitudes and frequencies for each tone.
Advanced Tone Detection (ATD)
ATD is a completely programmable tone detector. The ATD uses filters with a programmable bandwidth ranging from 36 to 582 Hz and programmable center frequencies with a resolution of 1 Hz. Debounce times, signal level thresholds, and signal-to-noise ratio (SNR) thresholds can also be configured. Up to 4 separate filters can be combined to detect a single multi-frequency tone. Up to 64 different filters can be defined and these can be combined to detect up to 128 unique tones.
Speech Detector
The speech detector detects the presence of speech, while ignoring other signals such as tones, modem signals and noise.
Generic FSK Modem (GFSK)
The iPline GFSK modem is an FSK modem capable of transmitting and receiving data at rates up to 1200 bits per sec. Any voice band FSK modem can be implemented, allowing all international variants of caller ID to be supported. We can provide configuration settings for caller ID, V.21, V.23, Bell 103 and Bell 202. We can also provide tools that will allow you to configure the GFSK modem as any voice band FSK modem.
Echo Cancellation
The iPline G.168 echo canceller can be used for VoIP, automatic speech recognition (ASR) barge-in and to enhance the performance of the DTMF detector, speech detector, and ATD in the presence of full duplex audio. The echo canceller supports tail lengths ranging from 1 ms up to 128 ms, configurable in 1 ms steps and is G.168 compliant over the full range of tail lengths. The echo canceller supports echo suppression (NLP) and comfort noise, both of which can be disabled. The double-talk, echo suppression and speech present thresholds are all configurable.
G3 FAX Modem
The iPline G3 FAX modem supports sending and receiving faxes at rates up to 14.4 Kbps. ITU-T V.21, V.27ter, V.29 and V.17 modems are supported. Together with the ITU-T T.30 host based application, iPline FAX supports:
• Error correction mode (ECM).
• Automatic detection of incoming calls.
• Header and footers added by senders and receivers.
• Subscriber ID.
• Third party viewers.
• T.4 and T.6 formats including:
- Metric and inch based pages size.
- All resolutions including 100×100, 200×200, 300×300 and 400×400.
- MH, MR and MMR coding.
Audio Conferencing
iPline audio conferencing has a latency of 0.375 ms, providing excellent audio quality and eliminating the need for echo cancellation in most situations. Conference configurations can be dynamically changed, allowing any conferee to be a member of any conference. Conferences can be bridged between DSPs, allowing conferences of virtually any size, from three conferees to hundreds of conferees, to be dynamically created. Input and output gain can be controlled and individual channels can be configured to operate in listener- or talker-only mode. With the support of echo cancellation, many useful configurations, such as coaching, can be created. With the DTMF detector, DTMF discrimination and clamping can be supported.
Gain Pad
The gain pad application can amplify or attenuate a pass-through signal with a latency of 0.375 ms.. The signal levels can be changed by any amount in the range of –40.0 to +24 dB in 0.1 dB increments.
MFR2
The DSP MFR2 application supports compelled R2 inter-register signaling. Both signal detection and signal transmission are supported. The protocol described in ITU-T Q.440 to Q.490 is supported as well as all the different variants of these standards. Both a manual mode and an automatic mode of operation are supported.
Channel Associated Signaling (CAS)
The CAS application supports both the detection and generation of CAS signals. The CAS application uses signal templates for both generation and detection, allowing custom defined signals, rather than just signaling transitions to be reported and generated.
Voice over IP (VoIP)
The DSP set of VoIP applications consists of a number of DSP applications, including G.711, G.726 and RTP, supporting RFC2833 for out of band DTMF signaling. Echo cancellation, tone generation and DTMF detection are also used.