estimation of voice transmission delay in tetra system - Telfor

ESTIMATION OF VOICE TRANSMISSIONDELAY IN TETRA SYSTEMMilorad Obradović 1 , Mirko Obradović 2 , Tomislav Nikolić 21 University of Novi Sad, Faculty of Engineering2 VLATACOM Ltd., Belgrade1. INTRODUCTIONThe important component parts of voice quality indigital trunking system from a user perspective are:ability to identify user (voice clarity); call set-up delayand voice transmission delay (not acceptable if toolarge). For example, in normal operating conditions, apolice user will consciously allow for call set-up delayand voice delay. However, in life threatening andstressful situations, a police officer will not consciouslyallow delay and will speak immediately into themicrophone and if no reply is received after a veryshort pause, will repeat the message. This lack ofconsciously allowing for delay will result in lack ofeffective communications at a time when it is mostneeded. If dissatisfaction is experienced in these areas,users will undoubtedly deem the system unacceptable.Voice quality is therefore the greatest challenge facingall TETRA [1] manufacturers.2. VOICE DELAY IN TRUNKING SYSTEMSIn the first analogue public safety trunking systemfounded that call set up times greater than 700ms wereunacceptable. Since those early experiences, it hasdetermined that call set up in a typical regional scenarioneeds to be sub 400 ms; this is in line with theSchengen requirement of 300 ms within one radio site.Similarly, it was founded that end to end voice packetdelays in excess of 500 ms were unacceptable fortalkgroup communications; and again bettering 400 mswas preferable.As well as analog trunking systems experience, it hasgained much experience through user reaction toTDMA integrated Dispatch, Telephony and Datasystem, first digital radio trunking system, called IDEN(Integrated Dispatch Enhanced Network) during itsdeployment in 1995, and subsequent acceptance in1996. Considerable effort was made to improve voiceclarity and delay, and has undoubtedly placed avaluable position in ensuring DIMETRA (DigitalMotorola Enhanced Trunked Radio) [3] learns fromthis experience. For example, from IDEN systemoperation, it was founded that end to end voice packetdelays in excess of 250 ms were unacceptable for fullduplex telephone interconnect.This again was rectified by considerable developmenteffort to optimise the digital voice coder, experiencethat has been used in the optimisation of the TETRAvoice coder on DIMETRA and end to end delayperformance. This experience and knowledge in thedevelopment of DIMETRA system has identified thatcall set-up times, end to end voice packet delay andvoice clarity are of prime importance for useracceptance. The TETRA standard imposes certaindelays in speech transmission because of itsfundamental design. This must be added to practicalconstraints in timing due to the underlying laws ofphysics.To achieve the user requirements for speech and calldelay, it was determined that any architecture thatrequired more than two switches to set up and carry acall would be unlikely to satisfy its end users; and thisdictated high performance peer to peer networkingdesign.The main reasons of delay which exists in the trunkingsystem, as TETRA is, are: the delay in speech coderand transmission delay.3. TETRA SPEECH CODERThe TETRA speech codec is based on the Code-Excited Linear Predictive (CELP) coding model [2]. Inthis model, a block of N speech samples issynthesized by filtering an appropriate innovationsequence from a codebook, scaled by a gain factor g c,through two time varying filters.A simplified diagram of this synthesis process, asimplemented in the TETRA codec, is shown in figure1.The first filter is a long-term prediction filter (pitchfilter) aiming at modeling the pseudo-periodicity in thespeech signal and the second is a short-term predictionfilter modeling the speech spectral envelope.

where T is the pitch delay and g p is the pitch gain. Thepitch synthesis filter is implemented as an adaptivecodebook, where for delays less than the sub-framelength the past excitation is repeated.The short-term synthesis filter is given by:1 1H ( z)= =, (2)pA(z)−i1+a z i∑i=1Figure1. The synthesis processThe long-term or pitch, synthesis filter is given by:where a i, i = 1,........, p, are the Linear Prediction (LP)parameters and p is the predictor order. In theTETRA codec p = 10.1B ( z )1=1 − g z p−T, (1)InputSpeechUnquantizedLPC infoLPC ANALYSISQUANTIZATION &INTERPOLATIONPastExcitationT0OPEN LOOPPITCHANALYSISPERCEPTUALWEIGHTINGTADAPTIVECODEBOOKgpgcLPC infoSHORT-TERMSYNTHESISFILTERkALGEBRAICCODEBOOKMSESEARCHPERCEPTUALWEIGHTINGGAIN VQGainsPitch delay (T)Codebook index (k)LPC infoMULTIPLEXDigitalOutputFigure 2. Block diagram of the TETRA speech coder

The TETRA encoder uses an analysis-by-synthesistechnique to determine the pitch and excitationcodebook parameters.In this analysis-by-synthesis technique, the syntheticspeech is computed for all candidate innovationsequences retaining the particular sequence thatproduces the output closer to the original signalaccording to a perceptually weighted distortionmeasure.The perceptual weighting filter de-emphasizes the errorat the formant regions of the speech spectrum and isgiven by:A(z)W ( z)= , (3)A(z / γ )where A(z) is the LP inverse filter and 0

layers, the physical and the link layer. It provides errordetection, but no error correction. Because of this, thenetwork buffering requirements are reduced, hence theincrease in throughput. Instead, the element of thesystem that guarantees the error-free end-to-endtransfer of frames are the endpoint devices, not thenetwork itself. This protocol processing, which is stillnecessary to guarantee the accurate delivery of thedata, is left to the higher layers inherent in thetransported data. The frame relay interfacespecification provides a signaling and data transfermechanism between endpoints and the network. Thisinterface allows communication bandwidth to beshared among multiple users, creating instantaneousbandwidth allocation on demand. Each frame (orpacket) contains header information that is used todetermine the routing of the data to the desireddestination. This enables each endpoint tocommunicate with multiple destinations via a singleaccess link to the network. Instead of fixed amounts ofbandwidth allocated to the resource, frame relay trafficreceives full bandwidth for short transaction bursts.Frame Relay uses the synchronous HDLC frameformat up to 4 kbytes in length. Each frame starts andends with a Flag character (7E Hex). The first 2 bytesof each frame following the flag contain theinformation required for multiplexing across the link,DLCI Field - Data Link Connection Identifiers. Thelast 2 bytes of the frame are always generated by aCyclic Redundancy Check (CRC) of the rest of thebytes between the flags, known as a frame checksequence (FCS).The rest of the frame contains the userdata. FRFPS, as bearer protocol, enables good end toend voice packet delay performance.Abstracts: This paper considers the estimation ofvoice transmission delay in digital trunking radiosystem TETRA. The main reasons of delay whichexists in the trunking system, as TETRA is, are: thedelay in speech coder and transmission delay. Thedelay of speech coder is about 80 ms. TETRA hasadvantage of Frame Relay Fast Packet Switching , sshort packet delay and minimal processing to meet thetiming requirement for real time voicecommunications. FRFPS, as bearer protocol, enablesgood end to end voice packet delay performance.Regarded to this protocol time delay in transmission isconsiderable less compared with speech coder delay.According to those principles, time delay is not largeand DIMETRA system has a satisfactory speechquality.REFERENCES[1] "Trans-European Trunking Radio (TETRA)",ETSI, 1996.[2] A. Gersho, "Advances in Speech and AudioCompression", Proceed. of IEEE, No. 6, June 1994.[3] "Dimetra-System Planer", MOTOROLA, 1998.[4] M. Obradović i dr., "Povezivanje baznih stanica usistem kod digitalnih mobilnih trankingsistema",JUKO CIRED,Herceg Novi, 2000.Regarded to this protocol time delay in transmission isconsiderable less compared with speech coder delayand call set-up is less than 300 ms.5. CONCLUSIONRecognising the importance of voice quality in digitalradio trunking system, it is necessary detailedevaluation of the following:• supplier experience in CELPC and network delayoptimisation,• system design configuration for optimum call setupperformance,• system design configuration for optimum end toend voice packet delay performance, and• chosen bearer circuit end to end delayperformance.According to those principles, time delay is not lageand DIMETRA system has a satisfactory speechquality.