October 17, 2005
VoIP in government organizations
Even though revenue and competition are not the driving forces that government bodies are subject to, they too need to cut costs. Convergence of voice and data provides government bodies with the opportunity to reduce costs and reap the benefits of enhanced functionality.
Government networks run on highly efficient TDM networks that use the same compression algorithms as VoIP networks. The difference lies in the fact that unlike VoIP compression, TDM voice compression is not accompanied by any other overheads. VoIP calls do have the advantage of bandwidth savings due to silence suppression. According to studies as much as 62% of a voice call is silent.
In a TDM setup, the bandwidth is dedicated to a call at the beginning of a call. The IP overhead that leads to increased bandwidth and the reduction in bandwidth due to silence suppression evens out the bandwidth consumption in VoIP calls and makes it comparable to that of TDM calls. Frame packing is an effective technique used to reduce the header size of VoIP packets. It involves loading several frames of voice packet into an IP packet.
The frames can be loaded onto a single packet in two ways. One method is to add several voice frames from the same voice call. A drawback of this method is the limitation in terms of the number of voice frames that can be added; too high a number can lead to delay. Another method is to use voice frames from different calls that are taking place at a given time. This technique allows frames from 60 different calls to be present in the same packet.
By minimizing the extra bandwidth used due to the IP headers, the advantage of silence suppression is felt more keenly. The bandwidth utilization of VoIP calls reduces to half that of compressed TDM calls. However, the commercially used standards such as H.323 and SIP do not offer the facility of frame packing. A typical characteristic of voice traffic is the small size of the multitude of packets generated. 33 small packets of 50 bytes each can be generated every second with VoIP calls. In contrast, a data packet has a maximum MTU size of 1500 bytes.
In order to maintain the QoS of voice services, voice traffic is differentiated from data traffic by DiffServ. STU-III and STU-IIB are standards used by U.S government agencies and NATO respectively, for the purpose of securing voice communications. The voice call is transferred as encrypted data over a modem. Modem calls do not support speech compression and therefore PCM (64 Kbps) is used for transferring them. Use of differential waveform coding (ADPCM) can reduce this to 32 Kbps but it impacts the modem transfer rate.
The problem of maintaining bandwidth efficiency as well as the security of the call can be solved by not using ADPCM and terminating the modem signal at the entry to the network. By extracting and transferring only the modulated data from the signals by means of a Secure Call Relay, bandwidth consumption can be managed. A buffer can help smooth jitter to a large extent with a minor delay in traffic. However, this is not possible with secure modem calls as delayed signals can lead to the termination side reconstructing the data sent by the originating side in an incorrect manner.
Error correction is utilized for prompt correction of corrupt VoIP packets without retransmission. Error correction is particularly useful with secure calls. A satellite hop can introduce a 250 ms one-way delay in a voice call and a half-minute round trip delay. Delay also increases the chances of the secure modem not remaining in sync. If satellites, such as INMARSAT, are being used in the government voice network, a BRI data interface can be used to connect the satellite transmitter to the network equipment.
Implementing VoIP in government networks can often mean having to deal with legacy systems that may have to be supported as well. The PBXs run on CAS protocols and applications such as MLPP may require support.
--
Did you enjoy this post?
