Dec 1, 2007

【好片推薦】不能說的祕密

昨天終於把CVOICE全面看完,該準備的補充資料也備齊了,該考的筆試也過了,所以終於可以喘口氣,原本準備在睡前把Cisco - Voice over IP Fundamentals看完之後就上床,不過因為看完之後才PM 11:00所以想說把之前堆積一堆電影看一下給自己放鬆一下。

挑來挑去就選了這一片好評不斷的電影 - 不能說的祕密。網路有一個好處,那就是可以根據大多數的人意見整合出一個很客觀的結果(雖然這個結果是來自於大多數人的主觀),所以我通常是根據這個指標來決定我下一部想看的影片。



"不能說的祕密" - 從這部影片的名稱看來其實可以猜到一二,一定跟愛情有關,因此原本有抱持著準備看連續劇的想法來看這部影片,不過片中的男女主角的感情發展說實在腳步進行的很快(不過這搭配上劇情來說也是合理,如果你到了另一個世界,只有一個異性能跟你溝通,只有這個人能夠知道你的存在,大概也會是這樣的情況)。

整部影片拍攝的感覺很好,漸漸脫離了港式江湖味,也少了台灣傳統國片的沈悶,但多了一點韓片的精緻及劇情細膩安排(也有人批評有抄襲之嫌,不過我想如果單以這部電影的劇情及結局安排來說,比韓片的悲情安排或刻意的喜劇收場來得特別)雖然角色不多,但是每個人的情緒及表現都溶入劇情中,雖然有些新人演技有點生澀,但無礙於整個劇情流暢的演進,更有趣地是常見的一點一滴漸漸揭開觀眾心中的問題之倒述法,令不少人在觀看電影的過程中猜測這是一部靈異片(就好似最近的一部西片-隱形人),基本上"不能說的祕密"應該要算是少見的科幻浪漫愛情國片。

看完這部片我個人最好奇的事情跟大部份人一樣那就是"結局"的合理化。由於只用了一張畢業照來帶過,因此給人很大的想像空間,這一點是我個人最喜愛的部份。在網路上討論最多的莫過於大家發表各自對於這樣結局的安排的劇情猜測,這個地方尤其讓人玩味的就是周杰倫出現在桂綸鎂的旁邊,但是許多人完全沒有注意到他的存在,因為…還蠻不像的,少了那個酷酷的感覺,變成一個靦腆的男學生。我看過周杰倫的頭文字D那部影片,說真的演技蠻爛的…但是從這部片中我感覺到他的改變,開始有電影男主角的感覺了,…至少話變多了雖然還是維持他的特色,每一段話都不長,但是多了許多的幽默的對白,這要感謝編劇的努力,不然這部片會少了一些笑聲。

其中的女主角小雨(桂綸鎂)及女配角(晴依)曾愷玹我覺得很棒,雖然對於曾愷玹來說戲份不多,但是曾愷玹的個人特質不錯,清新而亮麗,長相也很討喜,我想未來一定會漸漸發光。桂綸鎂就不用談了,很多人因為這部片喜歡上了桂綸鎂,包括她的下一部片-最遙遠的距離,已經有很多影迷拭目以待。我個人覺得桂綸鎂的演技一流,完全將女主角的特別表露無遺再加上個人的詮釋給予這個角色一些不同於劇本的特殊感覺,我想未來的星途不可限量,只要有好的劇本及導演的賞識,應該會有更棒的表現。

現在來談談這部片的結局,很多人覺得不合理的是為什麼周杰倫回到了過去,但是卻能出現在畢業照中,也就表示了在過去的時光中,周杰倫是可以被所有的人看得到的。個人的想法是劇情的安排,如果在琴房中正常的演奏琴譜,將可以直接跨越時空來到20年後的今天,但是將只會被20年後第一眼看到人的發現自己的存在;如果在琴房中以快速的節奏來彈琴,那麼就會直接回到20年前的今天(不過這邊留下了一個小小伏筆,似乎並不一定是同一天的同一個時間點,有可能比較早或是比較晚,這樣才有辦法讓周杰倫回到過去將桂綸鎂從無解的愛情中解救出來)。

「我閉上眼睛,是為了看清楚你。」這句話寫得很深清,更將影片中的重點深藏其中,唯有看過電影的觀眾才會心有戚戚焉,當電影散場之後各位再重新思考電影海報上這一句話,你會有一種大夢初醒的感覺。我的想法是如果從過去來到未來的規則中,是只能被第一眼看到的人所發現的,而其他人則完全視若無睹;但是如果是從未來回到過去的話,那麼就是一切正常,也就是將可以讓所有人的看見自己的存在。這樣的推論就可以讓周杰倫為何可以跟桂綸鎂一同出現在大家都可見的畢業照中了。只是比較沒有直接的證據可以佐證的是周杰倫回到的過去時間點到底是在桂綸鎂發現不可思議的琴譜之前或之後?:

  • 有人說是在桂綸鎂死前的那一天,看起來桂綸鎂並不認識周杰倫,事實上是因為桂綸鎂的調皮刻意假裝的…。不過這一點比較難以說明為何桂綸鎂會正常在學校教室中,因為桂綸鎂在氣喘發作當天之前五個月都沒有回到學校中(也沒有再去琴房跑到未來去找周杰倫,更不可能會乖乖地待在教室中寫東西。
  • 也有人說是在桂綸鎂發現琴譜之前,因此真的完全不認得周杰倫了,只是因為周杰倫的樣子讓她不自主地的發笑了。(這樣的推論是比較合理的。但是時間點的問題真的無法推論,因為如果可以透過彈琴的節奏來控制回到過去的時間點,可能會發生同一個時間點同時出現兩個一樣的人,所以…請自己合理化吧!)
  • 也有人說事實上周杰倫並來不及按下最後一琴鍵就已經死了,後面的劇情只是周杰倫死前最後的想像情節。(這也許是影迷最不希望的結局)
  • 個人比較喜歡的結局是周杰倫回到了桂綸鎂發現琴譜之後,已經認得周杰倫了,但是是在桂綸鎂撞見女配角跟周杰倫在琴房親嘴那一幕之前,所以桂綸鎂還是一貫地調皮刻意假裝不認得來捉弄周杰倫。(不過有點小小不合理啦,因為若真是如此,按理來說桂綸鎂的表現應該是驚喜不已才是,沒有想到周杰倫真的發現了這個祕密並且回到過去來找她,桂綸鎂要能控制住自己情緒還來捉弄周杰倫實屬不易…)



不論如何,這是一部成功的國片,尤其是片中的動畫場景真的已有國際性的水準,像是當桂綸鎂第一次演奏那份琴譜時,週遭事物的變化跟洋片的一些時間穿梭片段影像處理水準不相上下。雖然本片沒有像色戒中大卡司及名導演的加持,不過個人還是給予這部片極高的評價,只希望新生代的演員可以突破舊有國片惡劣的環境,重新讓國片受到國人及國際間的注視!

Nov 29, 2007

PCM vs ADPCM vs CS-ACELP vs LD-CELP

一旦我們的類比波形已經被數位化之後,我們可能想要透過編碼這些數位化波形來加以壓縮好節省廣域網路的頻寬。編碼和解碼這些波形的過程是由編解碼器(coder decoders),也被稱之為編解碼器(code decoder,CODEC)。讓我們來看看各種的編解碼器(CODECs)所使用的波形壓縮的一些形式︰

脈衝編碼調變(Pulse Code Modulation, PCM)
是將類比訊號轉換為數位訊號的一種技術,它並非實際地壓縮類比波形。相反的,脈衝編碼調變(PCM)取樣以及數位量化的動作並未進行任何的壓縮。G.711就是使用脈衝編碼調變(PCM)的編解碼器(CODEC)。

可適性差分脈衝碼調變(Adaptive Differentiated PCM,ADPCM)
使用一個差異化訊號(difference signal)。不將整個樣本加以編碼,可適性差分脈衝碼調變(ADPCM)可以將目前樣本與前一個樣本比較出來的差異傳送出去。
G.726就是一個可適性差分脈衝碼調變編解碼器(ADPCM CODEC)的例子。

代數碼激勵線形預測(Conjugate Structure Algebraic Code Excited Linear Predication,CS-ACELP)
動態建造基於語音模式的編碼登錄(codebook)。它會使用一個前看緩衝區(look ahead buffer)來查看是否下一個樣本與已存在於編碼登錄(codebook)中的圖案相配。如果它這樣做,那麼編碼登錄(codebook)位置就可以被傳送出去來取代實際的樣本。這個好處就是不需要傳送實際的聲音,我只要傳送給您那個聲音在您編碼登錄(codebook)中的位置,相較於傳送實際的聲音使用了相當少的頻寬。
G.729就是一個代數碼激勵線形預測編解碼器(CS-ACELP CODEC)的例子。

低時延碼激勵線性預測(Low-Delay Conjugate Excited Linear Predication,LD-CELP)
非常類似於代數碼激勵線形預測(CS-ACELP)。不過,低時延碼激勵線性預測(LD-CELP)使用了更小的編碼登錄(codebook),導致較少的延遲,但是它需要更多的頻寬。
G.728編解碼器(CODEC)是一個低時延碼激勵線性預測編解碼器(LD-CELP CODEC)的例子。

G.729所使用的前看緩衝區(look ahead buffer)的目的是為了在一個緩衝區中收集語音模式並且試圖將這些語音模式與已經存在於本地編碼登錄(codebook)中的模式來配對。事實上,在思科網路電話(VoIP環境)中,G.729是在廣域網路中最受歡迎傳送語音流量的編解碼器(CODEC),主要是因為它的高品質及低頻寬需求。為了傳送實際數位化的語音,G.729 只需要8 kbps,相較於G.711所需要的64 kbps頻寬。代數碼激勵線形預測(CS-ACELP)被設計用來把語音模式加以編碼。因此,其他音頻的來源(例如,來電等待音樂)與人類講話相比較,可能會經歷到更多的品質降低的情況。

您通常將會在本地區域網路環境中使用G.711(需64 kbps的頻寬來支付負荷語音流量)而在廣域網路環境中使用G.729(需8 kbps的頻寬來支付負荷語音流量)。所有種類的G.729需要8 kbps頻寬來傳送語音,以下是兩種G.729變型的差異之處:
  • G.729a使用一種較不複雜的演算法,保留處理器的資源伴隨著非常輕微的品質降低(quality degradation)。
  • G.729b支援語音活動偵測(voice activity detection,VAD)。

Note: 語音活動偵測(VAD)可以偵測談話什麼時候停止。在思科路由器上預設情況下,在250毫秒(ms)的寂靜(silence)之後(意即,4分之1秒),路由器停止傳送寂靜(silence),因此釋放出可用的頻寬。雖然語音活動偵測(VAD)所節省的頻寬數量基於語音模式變化而不同,但是通常可以節省百分之35的頻寬。

Understanding How Digital T1 CAS (Robbed Bit Signaling) Works in IOS Gateways

Channel Associated Signaling (CAS) is also referred to as Robbed Bit Signaling. In this type of signaling, the least significant bit of information in a T1 signal is "robbed" from the channels that carry voice and is used to transmit framing and clocking information. This is sometimes called "in-band" signaling. CAS is a method of signaling each traffic channel rather than having a dedicated signaling channel (like ISDN). In other words, the signaling for a particular traffic circuit is permanently associated with that circuit. The most common forms of CAS signaling are loopstart, groundstart, Equal Access North American (EANA), and E&M. In addition to receiving and placing calls, CAS signaling also processes the receipt of Dialed Number Identification Service (DNIS) and automatic number identification (ANI) information, which is used to support authentication and other functions.

Each T1 channel carries a sequence of frames. These frames consist of 192 bits and an additional bit designated as the framing bit, for a total of 193 bits per frame. Super Frame (SF) groups twelve of these 193 bit frames together and designates the framing bits of the even numbered frames as signaling bits. CAS looks specifically at every sixth frame for the timeslot's or channel's associated signaling information. These bits are commonly referred to as A- and B-bits. Extended super frame (ESF), due to grouping the frames in sets of twenty-four, has four signaling bits per channel or timeslot. These occur in frames 6, 12, 18, and 24 and are called the A-, B-, C-, and D-bits respectively.

The biggest disadvantage of CAS signaling is its use of user bandwidth in order to perform signaling functions.

Super Frame is an older framing standard for T1s. Also called D4 or D3/D4 framing. In the 1970s it replaced the original T1/D1 framing scheme of the 1960s in which the framing bit simply alternated.

In order to determine where each channel is located in the stream of data being received, each set of 24 channels is aligned in a frame. The frame is 192 bits long (8 * 24), and is terminated with a 193rd bit, the framing bit, which is used to find the end of the frame.

In order for the framing bit to be located by receiving equipment, a pattern is sent on this bit. Equipment will search for a bit which has the correct pattern, and will align its framing based on that bit. The pattern sent is 12 bits long, so every group of 12 frames is called a Super Frame. The pattern used in the 193rd bit is 1000 1101 1100.

Superframe remained in service in many places through the turn of the century, replaced by the improved Extended Super Frame of the 1980s in applications where its additional features were desired.

In telecommunication, an Extended Super Frame (ESF) is a T1 framing standard, sometimes called D5 framing, invented in the 1980s. It is preferred to its predecessor, Super Frame, because it includes a cyclic redundancy check (CRC) and bandwidth for a data link channel (used to pass out-of-band data between equipment.) It requires less frequent synchronization than the earlier superframe or D-4 format, and provides on-line, real-time testing of circuit capability and operating condition.

In ESF, a superframe is 24 frames long, and the 193rd bit of each frame is used in the following manner:

  • Frames 4, 8, 12, 16, 20, and 24 are used to send the framing pattern, 001011
  • Frames 1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21 and 23 are used for the data link (totalling half of all framing bits, or 4000 bits per second)
  • Frames 2, 6, 10, 14, 18, and 22 are used to pass the CRC total for each super frame.

Note: Less-frequent synchronization frees overhead bits for use in testing and monitoring.

CAS Signaling Types

Loopstart Signaling
Loopstart signaling is one of the simplest forms of CAS signaling. When a handset is picked up (the telephone goes off-hook), this action closes the circuit that draws current from the telephone company CO and indicates a change in status, which signals the CO to provide dial tone. An incoming call is signaled from the CO to the handset by sending a signal in a standard on/off pattern, which causes the telephone to ring.

A disadvantage of loopstart signaling is the inability to be notified upon a far-end disconnect or answer. For instance, a call is placed from a Cisco router configured for Foreign Exchange Station (FXS)-loopstart. When the remote end answers the call, there is no supervisory information sent to the Cisco router to relay this information. This is also true when the remote end disconnects the call.

Note: It is possible for answer supervision to be provided with loopstart connections if the network equipment can handle line-side answer supervision. Also, loopstart provides no incoming call channel seizure. Therefore a condition known as glare can arise, where both parties (Foreign Exchange Office [FXO] and FXS ) try to simultaneously place calls. Glare can be avoided when you configure the T1-CAS gateway's port selection order in such a way that the inbound and outbound calls are in reverse order. For example, if the inbound calls are sent by the provider on the FXO ports in the order of port 1, port 2, port 3 and port 4, then configure the Cisco CallManager Route Group to route outbound calls on those same ports in the order port 4, port 3, port 2 and port 1.

With loopstart signaling, the FXS side only uses the A-bit and the FXO side only uses the B-bit to communicate call information. The AB-bits are bi-directional. This state table defines this signaling information from the CPE's perspective (FXS).

This is the FXS-loopstart timing diagram.


On an incoming call (network -> CPE) this happens:

  1. The network toggles the B-bit to indicate ringing. This is a standard ringing pattern. For instance, 2 seconds on, 4 seconds off.
  2. CPE detects the ringing and off-hook states. A-bit goes from 0 to 1.

In an outgoing call (CPE -> network) this happens:

  1. CPE goes off-hook and A-bit goes from 0 to 1.
  2. The network provides dial tone. There is no signaling change.
  3. CPE sends digits (dual tone multifrequency (DTMF) in Cisco's case).

During a disconnect from the network, this occurs:

  1. CPE detects in-band that the call has dropped (someone says good-bye or a modem drops the carrier).
  2. CPE goes on-hook and A-bit goes from 1 to 0.

During a disconnect from the CPE, only step 2 occurs.

The Answer Supervision and Disconnect Supervision States are only seen when provided by the network.

Groundstart Signaling
Groundstart signaling is very similar to loopstart signaling in many regards. It works by using ground and current detectors that allow the network to indicate off-hook or seizure of an incoming call independent of the ringing signal and allow for positive recognition of connects and disconnects. For this reason, ground start signaling is typically used on trunk lines between PBXs and in businesses where call volume on loop start lines can result in glare.

The advantage of groundstart signaling over loopstart signaling is that it provides far-end disconnect supervision. Another advantage of groundstart signaling is the ability for incoming calls (network -> CPE) to seize the outgoing channel, thereby preventing a glare situation from occurring. This is done by using the A- and B- bit on the network side instead of just the B-bit. The A-bit is also used on the CPE side. However, the B-bit can also be involved, based on the switch's implementation. Typically the B-bit is ignored by the Telco. This is a state table that defines this signaling information from the CPE's perspective (FXS).

This is the FXS-groundstart timing diagram.



On an incoming call (network-> CPE) this happens:

  1. The network goes off-hook and the A-bit goes from 1 to 0 and rings the line by toggling the B-bit between 0 and 1.
  2. CPE detects the ringing and seizure and goes off-hook and the A-bit is set to 1.
  3. The network goes off-hook and the B-bit stops toggling. B-bit is now 1.

In an outgoing call (CPE -> network) this happens:

  1. CPE goes ground on ring and A-bit and B-bit are 0.
  2. The network goes off-hook and the A-bit goes from 1 to 0. The B-bit is set to 1.
  3. The CPE goes off-hook. The A-bit and the B-bit are 1.
  4. CPE detects a dialtone and sends digits.

During a disconnect from the network, this occurs:

  1. The network goes on-hook and the A-bit goes from 0 to 1.
  2. CPE goes on-hook and the A-bit goes from 1 to 0.

During a disconnect from the CPE, the above steps are reversed.

EandM Signaling
E&M Signaling is typically used for trunk lines. The signaling paths are known as the E-lead and the M-lead. Descriptions such as Ear and Mouth were adopted to help field personnel determine the direction of a signal in a wire. E&M connections from routers to telephone switches or to PBXs are preferable to FXS/FXO connections because E&M provides better answer and disconnect supervision.

E&M signaling has many advantages over the previous CAS signaling methods discussed in this document. It provides both disconnect and answer supervision as well as glare avoidance. E&M signaling is simple to understand and is the preferred choice when you use CAS.

This is the E&M Signaling diagram.



The three types of E&M Signaling that are supported on Cisco routers are:

  • Wink-start (FGB) - Used to notify the remote side that it can send the DNIS information.
  • Wink-start with wink-acknowledge or double-wink (FGD) - A second wink that is sent to acknowledge the receipt of the DNIS information.
  • Immediate start - Does not send any winks at all.

Note: FGD is the only variant of T1 CAS that supports ANI and Cisco supports it along with the FGD-EANA variant. In addition to FGD functionality, FGD-EANA provides certain call services, such as emergency (USA-911) calls. With FGD, the gateway supports the collection of ANI inbound only. With the use of FGD-EANA, a Cisco 5300 is able to send ANI information outbound as well as collecting it inbound. This latter capability requires the user of the fgd-eana signaling type in the ds0-group command, with ani-dnis option and calling-number outbound command in the POTS dial-peer. The calling-number outbound command is supported only on the Cisco 5300 as of Cisco IOS Software Release 12.1(3)T.

Therefore, on an incoming call (network-> CPE) this process happens:

  1. The network goes off-hook. The A-bit and B-bit equal 1.
  2. CPE sends wink. The A-bit and B-bit equal 1 for 200 ms. This only occurs when you use wink-start or wink-start with wink acknowledgement. Ignore this step for immediate start.
  3. The network sends DNIS information. This is done by sending inband tones which are decoded by the modem.
  4. CPE sends a wink acknowledgement. A-bit and B-bit equal 1 for 200 ms. This only occurs for wink-start with wink acknowledgement. Ignore this step for immediate start or wink-start.
  5. CPE goes off-hook when a call is answered. A-bit and B-bit equal 1.

In an outgoing call (CPE -> network) the same procedure occurs. However, the network just described is the CPE and vice-versa. This is because the signaling is symmetric.

During a disconnect from the network, this process occurs:

  1. The network goes on-hook. A-bit and B-bit equal 0.
  2. CPE goes on-hook. A-bit and B-bit equal 0.

During a disconnect from the CPE, these two steps are reversed.

Telephony Application Programming Interface(TAPI)

由微軟 (MICROSOFT) 與英特爾 (INTEL) 所共同推動的 API 規格, 可使 WINDOWS 應用程式直接或透過網路 (NETWORK) 來控制電話相關設備, 例如數據機 (MODEM), 頭戴話機 (HEADSET), 交換機 (PBX) 等. TAPI 的目標是要建立一種標準的規格來控制從簡單的撥號到電話中心 (CALL CENTER) 的控制, 在 NETWARE 上運作類似的規格稱為 TSAPI(Telephony Server Application Programming Interface)。

TAPI 3.0 是集合傳統式 PSTN 電話服務和 IP 電話服務的漸進式 API。IP 電話服務是新方興的技術,能夠在現有的區域網路、廣域網路 和 Internet 上融合聲音、資料和視訊。 TAPI 3.0 讓IP 電話服務在 Microsoft® Windows® 作業系統上成為可能﹔這方法可以簡單而普通的方法結合二或多部電腦,並存取這種連接上所包納的所有媒體資料流。

TAPI 3.0 支援標準的 H.323 會議和 IP 多點傳送會議。它使用 Windows 2000 作業系統的 Active Directory 服務來簡化公司內的部署,包含服務品質 (QoS) 支援,以提高會議品質,使網路易於管理。


TAPI 3.0 提供了簡單而普通的方法,能夠結合兩部或多部電腦,並存取這種結合所涵蓋的任何媒體資料流。它摘錄了呼叫控制功能,讓不同而看似不相容的傳輸通訊協定提供應用程式使用的公用介面。h當公司開始從昂貴而缺乏彈性的電路交換公用電話網路轉移至有智慧、彈性而且便宜的 IP 網路時,IP 電話服務能夠自若地面對爆炸性的成長。TAPI 3.0 整合多媒體資料流控制和傳統電話服務。此外,它是從 TAPI 2.1 API 到 COM 模式的改進,可用任何語言編寫 TAPI 應用程式,例如 C/C++ 或 Microsoft® Visual Basic® 。

Java Telephony Application Programming Interface(JTAPI)

談到JTAPI(Java Telephony Application Programming Interface),首先得瞭解什麼是CTI。

CTI(Computer Telephony Integration)就是電腦電話集成技術,它是目前國內正火的呼叫中心熱潮的核心技術。JTAPI主要是為CTI技術服務。JTAPI(Java Telephone API)則是一套專門為JAVA語言提供的與電話應用相關的程式介面,它定義了一組跨平臺、跨廠家的電話應用程式物件模型。使用JTAPI提供的物件,我們就可以簡單方便地用軟體實現各種CTI技術。

由於JTAPI的誕生是由若干知名電腦、通訊廠商(Sun, Lucent Technologies, Nortel, Novell, Intel, and IBM)聯合努力的結果,利用JTAPI編寫的CTI程式甚至可以操作若干種電話交換機,這些交換機包括Lucnet、Nortel等等廠家。

JTAPI的主要特點歸納如下:
1. 簡化CTI程式的編寫。
2. 提供一套可以擴展的框架結構,可以平滑的使Client/Server結構的程式過渡到Browser/Server結構。
3. 對已有的傳統CTI程式介面,如TSAPI、SunXTL、以及TAPI進行WEB方向的擴展。
4. 可以運行於任何JAVA可以運行的平臺。

利用以上優點,採用JTAPI技術搭建的呼叫中心就可以平滑的過渡到Internet時代。

目前JTAPI主要應用於呼叫中心領域,利用它還可以編寫包括自動撥號、語音郵件、傳真接收等各類軟體。特別在互聯網呼叫中心領域更是大有用武之地。比如Lucent 推出的ICC(Internet Call Center)就是一個典型的例子。整個ICC系統從技術上劃分,可以分為3部分:管理、CTI、工作流。三個部分都用JAVA開發,其中CTI部分使用JTAPI1.3。利用JAVA的優勢,ICC可以運行在NT、SALORIS等各種平臺之上。

Cisco Survivable Remote Site Telephony (SRST)

Cisco SRST Description

Cisco SRST provides Cisco CallManager with fallback support for Cisco IP phones that are attached to a Cisco router on your local network. Cisco SRST enables routers to provide call-handling support for Cisco IP phones when they lose connection to remote primary, secondary, or tertiary Cisco CallManager installations or when the WAN connection is down.

Cisco CallManager supports Cisco IP phones at remote sites attached to Cisco multiservice routers across the WAN. Prior to Cisco SRST, when the WAN connection between a router and the Cisco CallManager failed or when connectivity with Cisco CallManager was lost for some reason, Cisco IP phones on the network became unusable for the duration of the failure. Cisco SRST overcomes this problem and ensures that the Cisco IP phones offer continuous (although minimal) service by providing call-handling support for Cisco IP phones directly from the Cisco SRST router. The system automatically detects a failure and uses Simple Network Auto Provisioning (SNAP) technology to autoconfigure the branch office router to provide call processing for Cisco IP phones that are registered with the router. When the WAN link or connection to the primary Cisco CallManager is restored, call handling reverts back to the primary Cisco CallManager.

When Cisco IP phones lose contact with primary, secondary, and tertiary Cisco CallManagers, they must establish a connection to a local Cisco SRST router to sustain the call-processing capability necessary to place and receive calls. The Cisco IP phone retains the IP address of the local Cisco SRST router as a default router in the Network Configuration area of the Settings menu. The Settings menu supports a maximum of five default router entries; however, Cisco CallManager accommodates a maximum of three entries. When a secondary Cisco CallManager is not available on the network, the local Cisco SRST router's IP address is retained as the standby connection for Cisco CallManager during normal operation.

Note: Cisco CallManager fallback mode telephone service is available only to those Cisco IP phones that are supported by a Cisco SRST router. Other Cisco IP phones on the network remain out of service until they reestablish a connection with their primary, secondary, or tertiary Cisco CallManager.

Typically, it takes three times the keepalive period for a phone to discover that its connection to Cisco CallManager has failed. The default keepalive period is 30 seconds. If the phone has an active standby connection established with a Cisco SRST router, the fallback process takes 10 to 20 seconds after connection with Cisco CallManager is lost. An active standby connection to a Cisco SRST router exists only if the phone has the location of a single Cisco CallManager in its CallManager list. Otherwise, the phone activates a standby connection to its secondary Cisco CallManager.

Note: The time it takes for an IP phone to fallback to the SRST router can vary depending on the phone type. Phones such as the Cisco 7902, Cisco 7905, and Cisco 7912 can take approximately 2.5 minutes to fallback to SRST mode.

If a Cisco IP phone has multiple Cisco CallManagers in its CallManager list, it progresses through its list of secondary and tertiary Cisco CallManagers before attempting to connect with its local Cisco SRST router. Therefore, the time that passes before the Cisco IP phone eventually establishes a connection with the Cisco SRST router increases with each attempt to contact to a Cisco CallManager. Assuming that each attempt to connect to a Cisco CallManager takes about one minute, the Cisco IP phone in question could remain offline for three minutes or more following a WAN link failure.

Note: During a WAN connection failure, when Cisco SRST is enabled, Cisco IP phones display a message informing you that they are operating in Cisco CallManager fallback mode. The Cisco IP Phone 7960G and Cisco IP Phone 7940G display a "CM Fallback Service Operating" message, and the Cisco IP Phone 7910 displays a "CM Fallback Service" message when operating in Cisco CallManager fallback mode. When the Cisco CallManager is restored, the message goes away and full Cisco IP phone functionality is restored.

While in Cisco CallManager fallback mode, Cisco IP phones periodically attempt to reestablish a connection with Cisco CallManager at the central office. Generally the default time that Cisco IP phones wait before attempting to reestablish a connection to a remote Cisco CallManager is 120 seconds. The time can be changed in Cisco CallManager; see the "Device Pool Configuration Settings" chapter in the Cisco CallManager Administration Guide. A manual reboot can immediately reconnect Cisco IP phones to Cisco CallManager.

Once a connection is reestablished with Cisco CallManager, Cisco IP phones automatically cancel their registration with the Cisco SRST router. However, if a WAN link is unstable, Cisco IP phones can bounce between Cisco CallManager and Cisco SRST. A Cisco IP phone cannot reestablish a connection with the primary Cisco CallManager at the central office if it is currently engaged in an active call.

Figure 1 shows a branch office with several Cisco IP phones connected to a Cisco SRST router. The router provides connections to both a WAN link and the PSTN. The Cisco IP phones connect to their primary Cisco CallManager at the central office via this WAN link.

Figure 1 Branch Office Cisco IP Phones Connected to a Remote Central Cisco CallManager


Figure 2 shows the same branch office telephone network with the WAN connection down. In this situation, the Cisco IP phones use the Cisco SRST router as a fallback for their primary Cisco CallManager. The branch office Cisco IP phones are connected to the PSTN through the Cisco SRST router and are able to make and receive off-net calls.

Figure 2 Branch Office Cisco IP Phones Operating in SRST Mode


H.323 Gateways and SRST
On H.323 gateways, when the WAN link fails, active calls from Cisco IP phones to the PSTN are not maintained by default. Call preservation may work with the no h225 timeout keepalive command, but call preservation using the no h225 timeout keepalive command is not officially supported by Cisco Technical Support.

Under default configuration, the H.323 gateway maintains a keepalive signal with Cisco CallManager and terminates H.323-to-PSTN calls if the keepalive signal fails, for example if the WAN link fails. To disable this behavior and help preserve existing calls from local IP phones, you can use the no h225 timeout keepalive command. Disabling the keepalive mechanism only affects calls that will be torn down as a result of the loss of the H.225 keepalive signal.

MGCP Gateways and SRST
MGCP fallback is a different feature than SRST and, when configured as an individual feature, can be used by a PSTN gateway. To use SRST as your fallback mode on an MGCP gateway, SRST and MGCP fallback must both be configured on the same gateway. MGCP and SRST have had the capability to be configured on the same gateway since Cisco IOS Release 12.2(11)T.

To make outbound calls while in SRST mode on your MGCP gateway, two fallback commands must be configured on the MGCP gateway. These two commands allow SRST to assume control over the voice port and over call processing on the MGCP gateway. The two commands are ccm-manager fallback-mgcp and call application alternate. A complete configuration for these commands is shown in the "Enabling SRST on an MGCP Gateway" section.

Note: The commands listed above are ineffective unless both commands are configured. For instance, your configuration will not work if you only configure the ccm-manager fallback-mgcp command.


Enabling SRST on an MGCP Gateway
To use SRST as your fallback mode with an MGCP gateway, SRST and MGCP fallback must both be configured on the same gateway. The configuration below allows SRST to assume control over the voice port and over call processing on the MGCP gateway.

Note: The commands described in the configuration below are ineffective unless both commands are configured. For instance, your configuration will not work if you only configure the ccm-manager fallback-mgcp command.

SUMMARY STEPS
1. enable
2. configure terminal
3. ccm-manager fallback-mgcp
4. call application alternate [application-name]
5. exit

Skinny Client Control Protocol(SCCP)

SCCP is a proprietary terminal control protocol originally developed by Selsius Corporation. It is now owned and defined by Cisco Systems, Inc. as a messaging set between a skinny client and the Cisco CallManager. Examples of skinny clients include the Cisco 7900 series of IP phone such as the Cisco 7960, Cisco 7940 and the 802.11b wireless Cisco 7920, along with Cisco Unity voicemail server. Skinny is a lightweight protocol which allows for efficient communication with Cisco CallManager. CallManager acts as a signaling proxy for call events initiated over other common protocols such as H.323, SIP, ISDN and/or MGCP.

A skinny client uses TCP/IP to and from one or more Call Managers in a cluster. RTP/UDP/IP is used to and from a similar skinny client or H.323 terminal for the bearer traffic (real-time audio stream). SCCP is a stimulus-based protocol and is designed as a communications protocol for hardware endpoints and other embedded systems, with significant CPU and memory constraints.

Cisco acquired SCCP technology when it acquired Selsius Corporation in the late 1990s. As a remnant of the Selsius origin of the current Cisco IP phones, the default device name format for registered Cisco phones with CallManager is SEP -- as in Selsius Ethernet Phone -- followed by the MAC address.

Other companies like Symbol Technologies and SocketIP have implemented this protocol in VoIP Terminals (phones) and Media Gateway Controllers or Softswitches. Open Source implementation of SCCTP/Skinny exist and are now part of the Asterisk (PBX) system.

A company named IPBlue has created a software phone (soft phone) which uses SCCP for signaling, too. This phone in fact appears to the Cisco CallManager server as a 7960 hardware phone.

In addition, Cisco has come out with its own version of a skinny softphone called Cisco IP Communicator as well as SIP-based softphone called Cisco Unified Personal Communicator. Previously, Cisco had a JTAPI/CTI version of a softphone called Cisco IP Softphone.

Impairment / Calculated Planning Impairment Factor (ICPIF)

The ICPIF originated in the 1996 version of ITU-T recommendation G.113 "Transmission impairments," as part of the formula Icpif = Itot - A. ICPIF is actually an acronym for "(Impairment) Calculated Planning Impairment Factor," but should be taken to simply mean the "calculated planning impairment factor." The ICPIF attempts to quantify, for comparison and planning purposes, the key impairments to voice quality that are encountered in the network.

ICPIF Stands for “Impairment Calculated Planning Impairment Factor”. The ICPIF attempts to quantify, for comparison and planning purposes, the key impairments to voice quality that are encountered in the network. ICPIF values are expressed in a typical range of 5(very low impairment) to 55 (very high impairment). ICPIF values numerically less than 20 are generally considered “adequate”

Note: IP SLA uses a simplified formula which is also used by Cisco Gateways to calculate the ICPIF for received VoIP data streams.

The ICPIF is the sum of measured impairment factors (total impairments, or Itot) minus a user-defined access Advantage Factor (A) that is intended to represent the user's expectations, based on how the call was placed (for example, a mobile call versus a land-line call). In its expanded form, the full formula is expressed as:

Icpif = Io + Iq + Idte + Idd + Ie - A

where

•Io represents impairments caused by non-optimal loudness rating,

•Iq represents impairments caused by PCM quantizing distortion,

•Idte represents impairments caused by talker echo,

•Idd represents impairments caused by one-way transmission times (one-way delay),

•Ie represents impairments caused by equipment effects, such as the type of codec used for the call and packet loss, and

•A represents an access Advantage Factor (also called the user Expectation Factor) that compensates for the fact that users may accept some degradation in quality in return for ease of access.

ICPIF values are expressed in a typical range of 5 (very low impairment) to 55 (very high impairment). ICPIF values numerically less than 20 are generally considered "adequate." While intended to be an objective measure of voice quality, the ICPIF value is also used to predict the subjective effect of combinations of impairments. Table 1, taken from G.113 (02/96), shows how sample ICPIF values are expected to correspond to subjective quality judgement.

Table 1 Quality Levels as a Function of Total Impairment Factor ICPIF

Upper Limit for ICPIF  Speech Communication Quality
5               Very good
10              Good
20              Adequate
30              Limiting case
45              Exceptional limiting case
55              Customers likely to react strongly
               (complaints, change of network operator)

For further details on the ICPIF, see the 1996 version of the G.113 specification.

Note:The latest version of the ITU-T G.113 Recommendation (2001), no longer includes the ICPIF model. Instead, it refers implementers to G.107: "The Impairment Factor method, used by the E-model of ITU-T G.107, is now recommended. The earlier method that used Quantization Distortion Units is no longer recommended."

The full E-Model (also called the ITU-T Transmission Rating Model), expressed as R = Ro - Is - Id - Ie + A, provides the potential for more accurate measurements of call quality by refining the definitions of impairment factors (see the 2003 version of the G.107 for details). Though the ICPIF shares terms for impairments with the E-Model, the two models should not be confused.

The IP SLAs VoIP UDP Operation feature takes advantage of observed correspondences between the ICPIF, transmission rating factor R, and MOS values, but does not yet support the E-Model.

Nov 28, 2007

MOS vs PSQM vs PAMS vs PESQ

清晰度與語音訊號可接受的程度有關,舉例來說,收訊者是否能聽得懂對方所說的話,由聲音辨別發話者是誰或是由聲音感受發話者的感覺。

由於清晰度的因果關係並非線性,因此在許多與語音壓縮有關的數位技術中(例如MPEG-2),清晰度會有所謂的臨界效應(cliff effect);所謂臨界效應是指隨著訊號損失的增加,清晰度會逐漸變差,當清晰度變差到一個程度之後,收訊者便完全無法聽清楚,"cliff"的實際位置通常得靠實驗決定。

傳統上,是以平均意見指標(mean opinion score, MOS)來衡量清晰度;平均意見指標是將收訊的語音樣本,由一群收訊者依收聽到的通話品質分成5個等級:1代表最差、5代表最佳,4則是一般公眾電話網路系統的通話品質。由於MOS很難建立一個客觀標準,而且有實際執行上的困難,因此MOS無法作為長期評估的標準。

為了改善MOS的這些缺點,陸續有人希望藉由電腦輔助的方式,提出各種具有重複客觀性通話品質的評量方法。大部分的方式都是由收訊者的觀點,來比較以人類自然語音訊號作為語音樣本經過傳輸之後,接受訊號和原始訊號之間的差異。

目前,常用的清晰度評量方法有兩種,一種是由荷蘭KPN Research所發展的知覺通話質量測量(Perceptual Speech Quality Measurement, PSQM),現已成為ITU-T P.861標準;另一種是由大英國協的英國電訊所發展的知覺分析/測量系統(Perceptual Analysis/Measurement System, PAMS)

PSQMPAMS都使用自然語音(natural speech)或類語音樣本作為輸入訊號,通常選擇的語音樣本(speech sample)會經由語音傳輸路經傳送,語音傳輸路經在經過編碼、封包化(packetization)、傳輸和解碼的過程中,會造成各種不同程度的訊號損失。評量的方法是以接收的語音樣本訊號,和原本的訊號作為清晰度演算法的輸入訊號。典型測試所採用的語音樣本會包括,具有各種代表性的男性和女性聲音。

PSQM演算法是以0到6.5的數字來評量清晰度,數字越低代表通話品質越好。PSQM原本是設計用來評估和比較各種語音編碼(speech codecs)技術的優劣,而非點對點的(end-to-end)網路通話品質。但是,加強許多功能之後(稱為PSQM+)便可用來作為網路通話品質測試,在比較PSQMMOS的時候必須特別注意,PSQM與傳統MOS聽音品質間的關係並非線性。

經驗顯示,如果系統可以提供更多的其他服務,使用者可接受比目前公眾電話網路略差的通話品質。

PAMS會產生聽音品質指標(listening quality score)(Ylq)和聽音效應指標(listening effort )(Yle)兩種指標,它們都是由0~15編排,數字越高代表品質越好。和PSQM清晰度指標一樣,聽音品質指標主要是評量收訊者接收的語音訊號,與原本訊號之間的相似度。至於聽音效應指標則是不同的評量方式。

聽音效應指標主要是針對嚴重失真無法以聲音品質評估的訊號,因此聽音效應指標評估的是,收訊者必須花費多少心力才能聽懂嚴重失真的語音訊號所傳遞的訊息。

至於評估PSQMPAMS這類客觀語音品質評量演算法是否有效的方法,則是比對PSQMPAMS指標與MOS測試結果間,是否具有明顯的相關性。通常這些客觀演算法與主觀MOS評量法之間的相關性高達r>0.9。至於其他傳統的客觀評量方法,如噪訊比(signal-to-noise ratio)與MOS之間的相關性則很差,所以即使噪訊比很高也無法保證具有良好的通話品質。

PSQMPAMS的開發者KPN Research與英國電訊最近共同合作提出新的客觀語音品質評量ITU-T標準,稱為語音質量感知評估(Perceptual Evaluation of Speech quality, PESQ)這項技術結合PSQMPAMS兩種方法的優點—PSQM的聽覺模型(perceptual model)和PAMS的時間對位法(time-alignment routine),所以PESQ指標與MOS指標g之間的相關性將更高。PESQ分數範圍從1(最差)到4.5(最好),3.8代表一般傳統付費電話的可接受語音品質。

聲音變化偵測器(Voice Activity Detector, VAD)

為了對頻寬做最有效的利用,我們必須偵測聲音的變化,並且視需要來啟動或是停止封包的傳輸。聲音變化偵 測器所必須解決的最大問題,就是如何分辨說話的內容以及伴隨而來的背景雜訊,我們可以利用這項功能來節省網路的頻寬,因為在一般的通話過程中,幾乎一半的時間都沒有人說話。

Electrical Echo vs Acoustic Echo

在語音基頻訊號(basic level)時,回音基本上是自己聽到自己的反射聲音;就技術上的觀點來看,如果訊號的接受路徑(receive path)出現傳輸訊號時,便會產生回音,最常見且希望出現的回音方式是側音(sidetone),亦即在電話筒中聽到自己沒有任何時間延遲的聲音。而實際上,如果講電話時聽不到側音時,大部分人都會懷疑自己講的話對方是否聽得見。

回音本質上可分為電訊回音(electrical echo)及聲音回音(acoustic echo)。電訊回音主要是由於串音(crosstalk)或是阻抗不匹配(poor impedance matching)所造成的,至於聲波回音最常見的情況,則為:喇叭與遠端的麥克風產生相互作用而產生聲波回音(Acoustic echo)。

回音所造成的訊號干擾,會隨著回音的強度和回音的時間延遲而不同[如圖1所示],由於側音的時間延遲很短,因此除非強度很大才會造成明顯的干擾。但是由於語音封包網路的延遲時間,通常比傳統語音網路的延遲時間長了10倍以上,所以干擾會變得較為明顯。

就原理上來說,處理回音的方法中,最直接和較簡單的便是抑制回音;抑制回音的方法是在傳輸訊號時,關閉收訊路徑(receive path)。但是這個方法的問題是回音抑制電路(echo suppressor circuitry)必須能偵測發話者結束發話的時間,所以會造成類似時間延遲過長,所形成的半工(half-duplex)通訊的問題。

由於抑制回音的方法會有上述問題,因此較佳的方式為抵銷回音,由於回音抵銷的高性能數位訊號處理器價格越來越便宜,所以目前逐漸改用抵銷回音的技術,來取代回音抑制技術。當回音的延遲時間很短時,抵銷回音可發揮最大效果;故此回音抵銷技術,通常會與其他減少系統時間延遲的技術一併使用。

回音抑制器(Echo Suppressor)

一種裝置將其插裝到網路中,使信號在網路中只能伸單方向之傳輸,反方向之信號被抑制或消除,如此可消除同被信號。在電話通信網路中,如兩個中繼站之間的距離超過1850哩時,就需要加裝回音抑制器,以保持通話聲音之清晰。

標準的數據描述語言ASN.1 (Abstract Syntax Notation One) 簡介

ASN.1是一種用描述結構化客體的結構和內容的語言.

抽象語法定義:
ASN.1是描述在網絡上傳輸信息格式的標準方法。它有兩部分:描述信息內數據,數據類型及序列格式的是一部分;另一部分描述如何將各部分組成消息。它原來是作為X.409的一部分而開發的,來才自己獨立成為一個標準。ASN.1在OSI的ISO 8824/ITU X.208(說明語法)和ISO 8825/ITU X.209(說明基本編碼規則)規范。下面就是一個例子:

Report ::= SEQUENCE {
author OCTET STRING,
title OCTET STRING,
body OCTET STRING,
biblio Bibliography
}

在這個例子中,"Report"是由名字類型的信息組成的,而SEQUENCE表示消息是許多數據單元構成的,前三個數據單元的類型是OCTET STRING,而最一個數據類型則下面的ASN.1語法表示它的意義:

Bibliography ::= SEQUENCE {
author OCTET STRING
title OCTET STRING
publisher OCTET STRING
year OCTET STRING
}

(http://www.fanqiang.com)

MGCP Endpoint Identifiers

在Cisco CVOICE 5.0 P.3-105最下方的Example: Endpoint Identifiers的例子中只有文字但是缺少了附錄圖片,可能會讓各位在研讀內容時摸不著頭緒,所以我在網路上找到了這一段原來的文章內容(我猜是舊版的教材只是不小心圖片的部份被刪掉了…),請參考以下內容:

When interacting with a gateway, the call agent directs its commands to the gateway for the express purpose of managing an endpoint or a group of endpoints. An endpoint identifier, as its name suggests, provides a name for an endpoint.

Endpoint identifiers consist of two parts: a local name of the endpoint in the context of the gateway and the domain name of the gateway itself. The two parts are separated by an at sign (@). If the local part represents a hierarchy, the subparts of the hierarchy are separated by a slash. In Figure 6-35, the local ID might be representative of a particular gateway/circuit #, and the circuit # might in turn be representative of a circuit ID/channel #. In Figure 6-35, mgcp.gateway.cisco.com is the domain name, and t1toSJ/17 refers to channel 17 in the T1 to San Jose.

Figure 6-35. MGCP Endpoint Identifier

WiMax網路 獲企業用戶青睞

【本報紐約訊】新一代無線寬頻傳輸技術WiMax趨於成熟,網路服務供應商紛紛利用這項高科技為企業建置網路環境,企業用戶可用更低廉的成本、享受更快速的上網品質,並掙脫長久以來電話公司幾近於獨占的強勢掌控。
根據美聯社報導,近年來,網路服務供應商(ISP,Internet Service Provider)善用無線傳輸技術挑戰電話公司幾近獨占的地位,但效果有限,直到最近WiMax技術成熟,網路服務供應商挾著這項高科技的威力再度出征,終於獲得企業界青睞。

總部位於麻州的塔流公司(Towerstream)是近幾年才創立的固定/無線寬頻服務供應商,正積極在全美各地推廣WiMax企業網路服務,營運觸角涵蓋紐約、邁阿密、洛杉磯、芝加哥、西雅圖、舊金山、波士頓等地;以紐約為例,目前在曼哈頓一棟27層樓高建築物屋頂架設天線,透過天線可為一定範圍內的用戶提供資料傳輸服務。

塔流公司執行長湯普森(Jeff Thompson)站在天線所在地眺望紐約市景,視線範圍從格林威治村一路延伸到曼哈頓中城,視線所及的每一棟大樓都是塔流公司的潛在客戶;該公司所提供的寬頻服務傳輸速度最高可達每秒10億位元(1 gigabit),一般狀況下,傳輸速率為每秒2000萬位元(20 megabit),消費端的下載速度與藉助其他技術的傳輸速度差別不大,但資料上傳速度快很多,對企業來說是一大突破。

史普林新一代(Sprint Nextel)也計畫今年底前在芝加哥開始提供WiMax寬頻網路傳輸服務;不過,執行長佛西(Gary Forsee)10月初被迫下台為這項計畫埋下變數,為求因應,已另外規劃與清晰線路公司(Clearwire)合作。以開發企業用戶為主的無線網路服務供應商也躍躍欲試,德州的iBroad-band便是一例。

塔流公司的運作模式是,向相關單位取得高聳建築物屋頂使用權、而後架設天線、在鄰近區域開發企業用戶,然而,這套模式並非萬靈丹,有些業者已宣告陣亡;上選公司(Best Buy)轉投資的Speakeasy2004年便以西雅圖地標建築物Space Needle為基地提供服務,卻無疾而終,塔流公司今年出面接收,打算起死回生。

塔流第三季核心業務毛利率高達59%,但由於斥資擴張設備,出現160萬元虧損、營收170萬元。

Newscast公司目前是塔流的客戶,之前曾向電話公司租用標準低頻網路線T1,這家企業的副總蘇萊表示,T1很不牢靠,而且費用高,改用塔流的服務後,每月只需要付500元,是T1月租的一半,而且傳輸速度是T1的三倍。

新一代無線寬頻傳輸技術WiMax趨於成熟,網路服務供應商紛紛利用這項高科技為企業建置網路環境;剛成立不久的塔流公司便在全美各地推廣WiMax企業網路服務。圖為塔流公司執行長湯普森在紐約市曼哈頓下城一棟高樓屋頂上,展示該公司興建的WiMax天線。【美聯社】

Nov 26, 2007

Configuring Echo Cancellation

Echo cancellation is configured at the voice port level. It is enabled by default, and its characteristics are configurable. Echo cancellation commands are as follows:
  • echo-cancel enable—Enables cancellation of voice that is sent out through the interface and received back on the same interface. Sound that is received back in this manner is perceived by the listener as echo. Echo cancellation keeps a certain-sized sample of the outbound voice and calculates what that same signal looks like when it returns as an echo. Echo cancellation then attenuates the inbound signal by that amount to cancel the echo signal. If you disable echo cancellation, it will cause the remote side of a connection to hear echo. Because echo cancellation is an invasive process that can minimally degrade voice quality, you should disable this command if it is not needed. There is no echo path for a four-wire E&RM interface. The echo canceller should be disabled for this interface type.
  • echo-cancel coverage—Adjusts the coverage size of the echo canceller. This command enables cancellation of voice that is sent out through the interface and received back on the same interface within the configured amount of time. If the local loop (the distance from the interface to the connected equipment that is producing the echo) is longer, the configured value of this command should be extended.

    If you configure a longer value for this command, it takes the echo canceller longer to converge. In this case, the user might hear a slight echo when the connection is initially set up. If the configured value for this command is too short, the user might hear some echo for the duration of the call because the echo canceller is not canceling the longer-delay echoes. There is no echo or echo cancellation on the network side (for example, the non-POTS side of the connection).
  • non-linear—The function enabled by the non-linear command is also known as residual echo suppression. This command effectively creates a half-duplex voice path. If voice is present on the inbound path, then there is no signal on the outbound path. This command is associated with the echo canceller operation. The echo-cancel enable command must be enabled for the non-linear command to take effect. Use the non-linear command to shut off any signal if near-end speech is not detected.

    Enabling the non-linear command normally improves performance. However, some users encounter truncation of consonants at the ends of sentences when this command is enabled. This occurs when one person is speaking and the other person starts to speak before the first person finishes. Because the nonlinear cancellation allows speech in one direction only, it must switch directions on the fly. This might clip the end of the sentence spoken by the first person or the beginning of the sentence spoken by the second person.

QSIG(Q SIGnaling)

QSIG(Q sinnaling)是一種數位綜合服務網的協定,在專用數位交換網基於Q931標準的一種傳輸信號。Q信號被廣泛使用在IP網、虛擬私人網路、高速多功能企業網、教育網和企業機關網。

Q信號在不同的銷售商生產的設備組成的結點間傳遞時確保了Q931的基本功能。這些功能包括啟動(表示連接建立的信號)、處理信號(說明信號被目的端處理的信號)、響鈴警告(告訴互交方目的結點正在響鈴)、連接(返回呼叫端說明目的節電已經接收到信號)、釋放/完成(說明發送方或接受放已經中止了信號)。Q信號分兩層:BC層(基本層)和CF層(產生層)。BC層用於遮罩硬體差別,使信號在結點間透明傳輸。Q信號GF層為大型企業網、教育網、政府機關網提供了附加功能,如線性鑒定、呼叫中斷、呼叫分發、多媒體應用等。

Configuring Hookflash Relay on FXS/FXO Voice Ports

Introduction
When you integrate Voice over IP (VoIP) technologies to legacy private branch exchange (PBX) and public switched telephone networks (PSTNs), there is sometimes a need to pass a type of signaling known as 'hookflash'. A hookflash is a brief interruption in the loop current on loopstart trunks that the attached system does not interpret as a call disconnect.

Once the PBX or PSTN senses the hookflash, it generally puts the current call on hold and provides a secondary dial tone or access to other features such as transfer or call waiting access.

A hookflash is done by momentarily pressing down the cradle on a telephone. Some telephone handsets have a button called 'flash' or 'recall' that sends a 'timed loop break', or 'calibrated flash' which is a hookflash that has a precise timing.

Background Information
Many customers use a combination of FXS and FXO ports to extend telephone handsets across IP networks. They want to preserve features of the existing PBX, such as call forward, no answer to voice mail, and transfer/hold on the remote extensions. Earlier Cisco VoIP software did not provide full control to allow transparent integration. However, with the release of H.323 version 2 support in Cisco IOS Software Release 12.0.5T and later, it is now possible to detect and pass hookflash signaling across IP networks.

When the FXS port is configured for a long 'hookflash in' timer value (greater than 500 msec), users may complain that when they hang up and immediately pick up the handset, the call has not cleared. If the value is set too low, the hookflash may be interpreted as a hang-up, but a higher value means the handset has to be left hung-up for a longer period to clear the call. In some cases, cradle bounce can cause problems as well. As the handset is hung-up, the spring tension on the hook button causes multiple short breaks on the line known as cradle bounce. Careful tuning of the hookflash in timing value may be needed for best results. One possibility in such cases is to use handsets with a flash button that sends a hookflash of a specific period. The FXO port can be set to match this value and the FXO port then generates the outgoing hookflash. Many PBXes have a Class of Service (CoS) option called 'calibrated flash' or 'timed loop break' which allows them to recognize hookflashes of specific duration and to ignore other shorter or longer loop breaks. Such settings are helpful in eliminating false disconnects and generation of invalid hookflash signals to the PBX.

Configure
In this section, you are presented with the information to configure the features described in this document.

Note: To find additional information on the commands used in this document, use the Command Lookup Tool ( registered customers only) .

Configure PLAR OPX and Hookflash Relay
Use this procedure to configure private line, automatic ringdown (PLAR) Off-Premises extension (OPX), and hookflash relay.

  1. Configure the FXO port on the MainSite router as connection plar-opx.

    The OPX mode allows remote users on FXS ports to appear to a central PBX as a directly connected extension. When the FXO port detects a ring signal from the PBX, the router sends a VoIP call setup to the remote FXS port but it does not take the FXO port off-hook. As a result, the PBX only sees the call answer signal when the RemoteSide router FXS port is picked up. After the PBX reaches the no answer timeout (call rings out), then it can end the call, transfer the call to voice mail, or ring another extension/ring group. Without OPX mode, the FXO port immediately goes off-hook after it senses the ringing and the PBX is then unable to perform a call forward, no answer, or roll over to voice mail.
  2. The RemoteSite router must be configured to sense and then pass the hookflash signal on the FXS port.

    Since the hookflash is a momentary break in the loop current on the FXS port and cannot be sent as an audio signal, the router passes the hookflash signal via dual tone multifrequency (DTMF) relay as the '!' character. The router with the FXO port then sends a short loop break which the external device sees as a hookflash. To properly pass the hookflash signal, the VoIP dial peers need to be configured for dtmf-relay h245-signal.
  3. The physical port timers have to be adjusted to suit the characteristics of the handset on the FXS port and the duration of the hookflash loop break out of the FXO port as shown here:

    。The FXS voice port (RemoteSite router) uses the timing hookflash-in msec command where msec is the maximum value of a loop break (in milliseconds) from the telephone handset that is interpreted as a hookflash. A loop break greater than the configured value is regarded as a disconnect and the call is dropped. Any interval under this value causes the router to send the '!' character via the H.245-signal DTMF relay.

    。The FXO voice port (MainSite router) uses the timing hookflash-out msec command where msec is the duration of the outgoing loop break in milliseconds. When the router receives an H.245-signal DTMF relay signal, the FXO port generates a loop break for the configured interval.

E&M Signalling Interface

Introduction
This appendix provides additional information on the tie line signalling standards and the FastPAD's E&M interface. The material presented here supplements the information provided in Chapter 7.

Signalling Types
The FastPAD supports five E&M signalling standards (Types I through V) for PBX tie line interfaces. These conventions, as defined by AT&T specifications, are described below.

With each signalling type, the PBX supplies one signal, known as the M signal (for Mouth), and accepts one signal, known as the E signal (for Ear). Conversely, the tie line equipment (e.g., the FastPAD) accepts the M signal from the PBX and provides the E signal to the PBX. The M signal accepted by the tie line equipment at one end of a tie circuit becomes the E signal output by the remote tie line interface.

Each of the five types is illustrated in Figure G-1. The illustrations in this figure are abstracted from the specifications to show the essential components of the signalling circuitry. In this Figure G-1, the symbol V refers to battery voltage, which can be 25 Vdc to 65 Vdc, and is usually (nominally) -48 Vdc. Each of the illustrations in the figure show the PBX's E&M interface on the left, and the corresponding tie line equipment interface on the right.

Type I
With the Type I interface the tie line equipment generates the E signal to the PBX by grounding the E lead. The PBX detects the E signal by sensing the increase in current through a resistive load (this is indicated in the Figure G-1 by the unconnected node branching from the right side of the E resistor). Similarly, the PBX generates the M signal by sourcing a current to the tie line equipment, which detects it via a resistive load.

The Type I interface requires that the PBX and tie line equipment share a common signalling ground reference. This can be achieved by connecting signal ground from the PBX to the SG lead (pin 8) of the RJ45 connector.

Type II
The Type II interface requires no common ground; instead, each of the two signals has its own return. For the E signal, the tie line equipment permits current to flow from the PBX; the current returns to the PBX's ground reference. Similarly, the PBX closes a path for current to generate the M signal to the tie line equipment.

Type III
A variation of Type II, Type III uses the SG lead to provide common ground. With this configuration, the PBX drops the M signal by grounding it, rather than by opening a current loop.

Type IV
Type IV is symmetric and requires no common ground. Each side closes a current loop to signal; the flow of current is detected via a resistive load to indicate the presence of the signal.

Type V
Type V is a simplified version of Type IV. This is also a symmetric interface, using only two wires. Type V requires a common ground between the PBX and the tie line equipment; this is provided via the SG leads.


Figure G-1: E&M Signalling Types



E&M Interface Types and Wiring Arrangement
There are five different E&M interface types or models named Type I, II, III, IV, and V (Type IV is not supported on Cisco platforms). Each type has a different wiring arrangement, hence a different approach to transmit E&M supervision signaling (on-hook / off-hook signaling). The signaling side sends its on-hook/off-hook signal over the E-lead. The trunking side sends the on-hook/off-hook over the M-lead.
For more information and pinout diagrams of E&M types, refer to Understanding and Troubleshooting Analog E&M Interface Types and Wiring Arrangements.

  • E&M Type I—This is the most common interface in North America.
    。Type I uses two leads for supervisor signaling: E, and M.
    。During inactivity, the E-lead is open and the M-lead is connected to the ground.
    。The PBX (that acts as trunk circuit side) connects the M-lead to the battery in order to indicate the off-hook condition.
    。The Cisco router/gateway (signaling unit) connects the E-lead to the ground in order to indicate the off-hook condition.
  • E&M Type II—Two signaling nodes can be connected back-to-back.
    。Type II uses four leads for supervision signaling: E, M, SB, and SG.
    。During inactivity both the E-lead and M-lead are open.
    。The PBX (that acts as trunk circuit side) connects the M-lead to the signal battery (SB) lead connected to the battery of the signaling side in order to indicate the off-hook condition.
    。The Cisco router / gateway (signaling unit) connects the E-lead to the signal ground (SG) lead connected to the ground of the trunk circuit side in order to indicate the off-hook condition.
  • E&M Type III—This is not commonly used in modern systems.
    。Type III uses four leads for supervision signaling: E, M, SB, and SG.
    。During inactivity, the E-lead is open and the M-lead is set to the ground connected to the SG lead of the signaling side.
    。The PBX (that acts as trunk circuit side) disconnects the M-lead from the SG lead and connects it to the SB lead of the signaling side in order to indicate the off-hook condition.
    。The Cisco router / gateway (signaling unit) connects the E-lead to the ground in order to indicate the off-hook condition.
  • E&M Type IV—This is not supported by Cisco routers / gateways.
  • E&M Type V—Type V is symmetrical and allows two signaling nodes to be connected back-to-back.
    。This is the most common interface type used outside of North America.
    。Type V uses two leads for supervisor signaling: E, and M.
    。During inactivity the E-lead and M-lead are open.
    。The PBX ( that acts as trunk circuit side) connects the M-lead to the ground in order to indicate the off-hook condition.
    。The Cisco router / gateway (signaling unit) connects the E-lead to the ground in order to indicate off-hook condition.

Application Examples
In examples below the term "attached device" refers to tie line equipment such as the FastPAD.

E&M Type I

Idle Condition. The E lead is biased by the PBX at -48 Vdc. The M lead is biased at 0 Vdc by the attached device (FastPAD).


PBX Initiated Call. The PBX supplies -48 Vdc to the M lead, signalling to the attached device that it wants a connection. The attached device grounds the E lead, signalling the response.


Attached Device Initiated Call. The attached device grounds the E lead, signalling to the PBX that it wants a connection. The PBX supplies -48 Vdc to the M lead, signalling the response.


PBX Initiated Disconnect. The PBX initiates disconnection by opening the M lead. The M lead is pulled to 0 Vdc, as biased by the attached device. In response, the attached device opens the E lead, which is pulled to -48 Vdc, as biased by the PBX.


Attached Device Initiated Disconnect. The attached device initiates the disconnect by opening the E lead. The E lead is pulled to -48 Vdc as biased by the PBX. In response, the PBX opens the M lead. The M lead, which is pulled to 0 Vdc, as biased by the attached device.

E&M Type II

Idle Condition. The E lead is biased by the PBX at -48 Vdc. The M lead is biased at -48 Vdc by the attached device (FastPAD).


PBX Initiated Call. The PBX grounds the M lead, signalling to the attached device that it wants a connection. The attached device grounds the E lead, signalling the response.


Attached Device Initiated Call. The attached device grounds the E lead, signalling to the PBX that it wants a connection. The PBX grounds the M lead, signalling the response.


PBX Initiated Disconnect. The PBX initiates disconnection by opening the M lead. The M lead is pulled to -48 Vdc, as biased by the attached device. In response, the attached device opens the E lead, which is pulled to -48 Vdc, as biased by the PBX.


Attached Device Initiated Disconnect. The attached device initiates the disconnect by opening the E lead. The E lead is pulled to -48 Vdc as biased by the PBX. In response, the PBX opens the M lead. The M lead, which is pulled to -48 Vdc, as biased by the attached device.

E&M Type V

Idle Condition. The E lead is biased by the PBX at -48 Vdc. The M lead is biased at -48 Vdc by the attached device (FastPAD).


PBX Initiated Call. The PBX grounds the M lead, signalling to the attached device that it wants a connection. The attached device grounds the E lead, signalling the response.


Attached Device Initiated Call. The attached device grounds the E lead, signalling to the PBX that it wants a connection. The PBX grounds the M lead, signalling the response.


PBX Initiated Disconnect. The PBX initiates disconnection by opening the M lead. The M lead is pulled to -48 Vdc, as biased by the attached device. In response, the attached device opens the E lead, which is pulled to -48 Vdc, as biased by the PBX.


Attached Device Initiated Disconnect. The attached device initiates the disconnect by opening the E lead. The E lead is pulled to -48 Vdc as biased by the PBX. In response, the PBX opens the M lead. The M lead, which is pulled to -48 Vdc, as biased by the attached device.

FastPAD E&M Interface
The FastPAD's E&M interface is designed to connect with that of a PBX tie line port, and provide appropriate end-to-end signalling support for a variety of applications. The following paragraphs describe this interface in detail.


FastPAD Circuits and E&M Signalling. The FastPAD generates the E signal to the PBX in response to an inbound signal at the remote FastPAD. That signal depends on the application of the remote unit. The local FastPAD will generate the E signal in the following applications:


The remote FastPAD is configured for E&M signalling, and detects an active M signal, The remote unit uses the FastPAD OPX option, and detects an off hook condition on its two-wire loop-start or ground-start circuit. The remote unit uses the FastPAD SLT option, and detects a ring signal on its two-wire loop-start or ground-start circuit.