Extended-rtPS Algorithm for VoIP Services
in IEEE 802.16 systems
Howon Lee
∗, Taesoo Kwon
†and Dong-Ho Cho
‡Department of Electrical Engineering and Computer Science
Korea Advanced Institute of Science and Technology (KAIST)
373-1, Guseong-dong, Yuseong-gu, Daejeon, Korea
Telephone: +82-42-869-3467, Fax: +82-42-867-0550
Email:
{
∗hwlee,
†tskwon80
}@comis.kaist.ac.kr,
‡[email protected]
Abstract— There are several scheduling algorithms for Voice
over IP (VoIP) services in IEEE 802.16 systems, such as un-solicited grant service (UGS), real-time polling service (rtPS), UGS with Activity Detection (UGS-AD), and Lee’s algorithm using Grant-Me bit of the generic MAC header. However, these algorithms have some problems of a waste of uplink resources, additional access delay, and MAC overhead for supporting VoIP services with variable data rates and silence suppression. To solve these problems, we propose a novel uplink scheduling algorithm (Extended-rtPS) for the VoIP services in IEEE 802.16 systems. Through the performance analysis and simulation results of resource utilization, VoIP capacity, total throughput, and packet transmission delay, we show that our proposed algorithm can solve the problems of the conventional algorithms, and has the best performance among these algorithms. In addition, with simulation results of packet transmission delay, we prove that our proposed algorithm can support more 74%, 24%, and 9% voice users compared with the UGS, rtPS, and UGS-AD (Lee’s) algorithms, respectively.
I. INTRODUCTION
T
HE IEEE 802.16 standard is designed to satisfy various demands for higher capacity, higher data rate, and more advanced multimedia services [1], [2]. This standard has many advantages, such as rapid deployment, high speed data rate, high scalability, multimedia services, and lower maintenance and upgrade costs. Especially, the IEEE 802.16’s Task Group (TG) d/e systems have proposed to make up for the weak points of 16a system, and support additional functionalities, such as mobility, hybrid automatic repeat request (HARQ), band adaptive modulation coding (Band-AMC) scheme, and so on.There are four uplink scheduling algorithms to support variable requirements of QoS in IEEE 802.16 systems, such as UGS, rtPS, non-rtPS (nrtPS), and best effort service (BE). However, these algorithms are not suitable for VoIP services with variable data rates and silence suppression in IEEE 802.16 systems.
In addition, there are some previously proposed algorithms for the VoIP services, such as UGS-AD in the DOCSIS system, and Lee’s algorithm [3]- [5]. But, these algorithms also have problems for supporting VoIP services with variable data rates and silence suppression. That is, these algorithms cannot
This research was supported in part by University IT Research Center Project.
support voice codecs with variable data rates well, such as enhanced variable rate codec (EVRC) [6]. The UGS-AD and Lee’s algorithms can follow voice codec with only two data rates (on-off). Therefore, we propose a novel uplink scheduling algorithm for the VoIP services, ertPS. This algorithm is recently proposed and adopted in IEEE 802.16e standard by us [7]. On the contrary, although the UGS-AD algorithm is also proposed, this algorithm is rejected in this standard [2], [4].
In this paper, for performance analysis of conventional algorithms and our proposed algorithm (UGS, rtPS, UGS-AD, Lee’s, ertPS), we utilize EVRC with variable data rates and silence suppression [6]. The frame duration of EVRC (Tvc)
is20ms, and the voice activity factor is 0.403 with 29% full rate (Rate 1, p1), 4% half rate (Rate 1/2, p2), 7% quarter rate (Rate 1/4,p3), and 60% eighth rate (Rate 1/8,p4). Also, the sizes of voice packets in each rate, (L1, L2, L3, L4) are 171, 80, 40 and 16 bits. Full, half and quarter rates are included in talk-spurt (on) duration, and eighth rate is included in silence (off) duration. When using a voice codec with a voice activity detector (VAD) or silence detector (SD), the voice user can know his voice status in the higher layer. This higher layer information can be known in the MAC layer by using primitives of Convergence Sublayer (CS layer) in IEEE 802.16e systems [2], [8].
The remainder of this paper is organized as follows: In Section II, we introduce and discuss conventional scheduling algorithms, such as the UGS, rtPS, UGS-AD, Lee’s, nrtPS, and BE algorithms. In Section III, we propose a novel scheduling algorithm to solve the problems of the conventional algo-rithms. Also, in Section IV and V, we analyze and simulate the performance of resource utilization, VoIP capacity, throughput, and packet transmission delay of the conventional and pro-posed algorithms. Finally, in Section VI, we make conclusions. II. CONVENTIONALUPLINKSCHEDULINGALGORITHMS A. UGS Algorithm
The UGS algorithm is designed to support real-time service flows that generate fixed-size data packets periodically [1]. In this algorithm, the base station (BS) periodically assigns fixed-size grants to the voice user, and these fixed-fixed-size grants are suf-ficient to send voice data packets generated by the maximum
Time
Resource
ON OFF
Dot line : Assigned resources Solid line : Used resources
Data rate increment Data rate decrement Rate 1
Rate 1/2
Rate 1/4
Rate 1/8
Fig. 1. Operation of UGS algorithm
data rate of EVRC. Thus, this algorithm can minimize MAC overhead and uplink access delay caused by the bandwidth request process of the user to send voice packets.
However, this algorithm has only a small capacity for VoIP services. Generally, voice users do not always have voice packets with the same size, because they have variable data rates and silence suppression [6]. In this algorithm, since the BS always assigns fixed-size grants that are sufficient to send voice packets generated by the maximum data rate of EVRC, it causes a waste of many uplink resources, as shown in Fig. 1. In this figure, a dot line and a solid line show the amount of assigned uplink resources by the BS and the amount of used uplink resources by the user, respectively. The blank regions represent the waste of uplink resources. In other words, in the UGS algorithm, since the BS always allocates the same amount of uplink resources to each user regardless of his voice status, it causes the waste of a lot of uplink resources.
B. rtPS Algorithm
The rtPS algorithm is designed to support real-time service flows that generate variable size data packets periodically [1]. The BS assigns uplink resources that are sufficient for unicast bandwidth requests to the voice user. Generally, this process is called a bandwidth request process, or polling process. Be-cause this algorithm always uses a bandwidth request process for suitable size grants, it transports data more efficiently than the UGS algorithm. However, this bandwidth request process always causes MAC overhead and additional access delay. Hence, the rtPS algorithm has larger MAC overhead and access delay than the UGS, UGS-AD, Lee’s and our proposed algorithms.
In Fig. 2, since the user requests exact amount of uplink resources for transmitting his voice packets, the dot line and the solid line are nearly the same. In this paper, to avoid the polling process in a silence duration, we assume that minimum polling size is the size of voice packet generated by the minimum data rate of the voice codec, except the UGS algorithm. Thus, there is no polling process in the silence duration of the users.
In the rtPS algorithm, the user can use the piggyback requests of the grant management subheader for VoIP services.
Time
Resource
ON OFF
Dot line : Assigned resources Solid line : Used resources
Data rate increment Data rate decrement Rate 1
Rate 1/2
Rate 1/4
Rate 1/8 Bandwidth request header
Fig. 2. Operation of rtPS algorithm
But, because VoIP services are delay-sensitive, the usage of piggyback requests is not a desirable method for VoIP services. By precise negotiation of the polling period in the initialization process, the use of piggyback requests may be avoided for VoIP services.
C. UGS-AD Algorithm
The UGS-AD algorithm is designed to support real-time service flows that generate fixed size data packets on a semi-periodic basis [3], [4]. This is a combined algorithm of the UGS and rtPS algorithms. This algorithm has two scheduling modes (UGS, rtPS), and can switch these modes according to the status of voice users. If the VoIP services were initiated, this algorithm firstly starts as the rtPS mode. In case of the rtPS mode, if the voice user requests bandwidth size as zero byte, the BS maintains its mode. However, if the user requests another bandwidth size (non-zero bytes), the BS has to switch its mode to the UGS. In case of the UGS mode, the BS operates by contraries.
As shown in Fig. 3, in the UGS-AD algorithm, since the BS cannot follow the half and quarter rates of EVRC, the waste of uplink resources would occur. In case of the eight rate, this waste is not caused, because the BS switches its mode to the rtPS. When a data rate of the user is changed from the quarter rate to the eighth rate, for notifying his status, the user utilizes grant management subheader with a bandwidth request of zero byte [2]. This process does not cause MAC overhead, since the user uses remained uplink resources. In case that the data rate of the voice user is increased to the full rate, he should transmit bandwidth request header with a bandwidth request of non-zero bytes [2]. In this case, at first, the BS allocates enough uplink resources to the user for sending first delayed voice packet and second voice packet generated by the full rate of EVRC. Then, the BS periodically assigns uplink resources according to the general operation of the UGS mode.
The UGS-AD algorithm can solve the problems caused by the UGS algorithm (waste of uplink resources) and the rtPS algorithm (MAC overhead and access delay), in case that the voice users use voice codecs with only two data rates (on-off). However, in case of EVRC with variable data rates and
Time
Re
so
urce
ON OFF
Dot line : Assigned resources Solid line : Used resources
Data rate increment Data rate decrement Rate 1
Rate 1/2
Rate 1/4
Grant Management subheader
Rate 1/8 Bandwidth request header
UGS mode rtPS mode
Fig. 3. Operation of UGS-AD algorithm
silence suppression, this algorithm cannot solve the problems perfectly. In this case, the waste of uplink resources occurs in talk-spurt (on) duration of the voice users.
D. Lee’s Algorithm
Similar to the UGS-AD algorithm, Lee’s algorithm partially can solve the problems of the UGS and rtPS algorithms, in that the BS basically assigns uplink resources to voice users by considering only on-off transitions of them. In this algorithm, because the voice user has to inform the BS of his voice state transitions, it requires a method for relaying his voice status information. Therefore, since the conventional generic MAC header of IEEE 802.16 systems has two reserved bits for other additive operations, Lee’s algorithm uses one reserved bit in this header for a method to inform the BS of the user’s voice state transitions [1], [2]. This one reserved bit is defined as a Grant-Me (GM) bit in this algorithm.
When the voice state of the user is ‘on’, he sets the GM bit to ‘1’, otherwise he sets the GM bit to ‘0’. The voice user can inform the BS of his voice state transitions effectively without MAC overhead, because it conveys the voice status information using the conventional generic MAC header. Detailed operation of this algorithm according to the GM bit is as follows.
If the GM bit is ‘0’: The BS assigns the minimum grant
size to the voice user. This minimum size is sufficient in case that the user informs the BS of his voice state transitions. When the GM bit is changed, from ‘1’ into ‘0’, the BS once assigns maximum grant size to the user whose voice state is ‘off’, as shown in Fig. 4. Hence, it causes a little waste of uplink resources, which could be negligible.
If the GM bit is ‘1’: The BS assigns the maximum grant
size to the voice user. That size is sufficient to send voice data packets. In case that the GM bit is changed, from ‘0’ into ‘1’, the BS once assigns minimum grant size to the user whose
Time
Resource
ON OFF
Dot line : Assigned resources Solid line : Used resources
Data rate increment Data rate decrement Rate 1
Rate 1/2
Rate 1/4
Rate 1/8 Bandwidth request header
Fig. 4. Operation of Lee’s algorithm
voice state is ‘on’. Thus, the voice user cannot send the voice packet using this grant size. In this case, the user can send this voice packet by using piggyback requests of the grant management subheader or bandwidth requests of bandwidth request header, as shown in Fig. 4.
E. nrtPS and BE algorithms
The nrtPS and BE algorithms are designed to support non-realtime service flows, such as HyperText Transfer Protocol (HTTP) and File Transfer Protocol (FTP) [1], [2]. In other words, these algorithms are not proper to use for VoIP services with variable data rates and silence suppression. Thus, in this paper, we did not consider these algorithms, since we are focused on realtime VoIP services.
III. PROPOSEDALGORITHM
To solve the problems of the UGS, rtPS, UGS-AD, and Lee’s algorithms, such as the waste of uplink resources, MAC overhead, and additional access delay, we propose a novel uplink scheduling algorithm for VoIP services with variable data rates and silence suppression in IEEE 802.16 systems. Basically, in order to solve these problems, our proposed algorithm allocates uplink resources according to all status of the voice users without MAC overhead. Here, we describe the detailed operation of our proposed algorithm.
Firstly, the voice user informs the BS of his voice status information using grant management subheader in case that the size of a voice data packet is decreased [2]. The user requests the bandwidth for sending the voice packets using extended PBR (piggyback request) bits of grant management subheader. Since the user uses remained uplink resources assigned to him, there is no waste of uplink resources. In our proposed algorithm, to distinguish these extended PBR bits with general PBR bits, the user sets the MSB of PBR bits to 1. In this case, the BS assigns uplink resources according to the requested size periodically, until the voice user requests another size of the bandwidth.
Secondly, the voice user informs the BS of his voice status information using bandwidth request header in case that the size of a voice data packet is increased [2]. The user
Time
Resource
ON OFF
Grant Management subheader Dot line : Assigned resources
Solid line : Used resources Bandwidth request header
Data rate increment Data rate decrement Rate 1
Rate 1/2
Rate 1/4
Rate 1/8
Fig. 5. Operation of proposed algorithm
requests the bandwidth for sending the voice packets using BR (bandwidth request) bits of bandwidth request header. In the same way as the case of the data rate decrement, to distinguish these BR bits with general BR bits, the user sets the MSB of BR bits to 1. Then, the BS also assigns uplink resources according to the requested size periodically, until the user requests another size of the bandwidth. In case of the data rate increment, the BS shall provide the first bandwidth allocation to the next MAC frame after this bandwidth request process. The second bandwidth allocation is done after the bandwidth allocation based on the time which the BS allocated the bandwidth used for the bandwidth request process, as shown in Fig. 5.
In summary, in case of VoIP services with variable data rates and silence suppression using this algorithm, the BS recognizes grant management subheader and bandwidth re-quest header especially. In this algorithm, if the user rere-quests the bandwidth for sending the voice packets, then the BS shall change its polling size according to the bandwidth size requested by the user, and keeps its changed polling size until the user sends another requests. In other words, the BS may not change its polling size without any requests from the voice users. Since the proposed algorithm can follow all data rates of the voice users, the BS can obtain better data transport efficiency compared with the UGS, rtPS, UGS-AD and Lee’s algorithms.
IV. PERFORMANCEANALYSIS A. Resource Utilization
In case that the total number of voice users is NT ot, the
numbers of users in 1, 1/2, 1/4, and 1/8 rates (N1, N2, N3,
N4) are obtained by N1 = NT ot× p1, N2 = NT ot × p2,
N3 = NT ot× p3, and N4 = NT ot× p4. By using N1, N2,
N3, andN4, we can calculate the numbers of required uplink
resources in the UGS, rtPS, UGS-AD, Lee’s, and our proposed algorithms. The numbers of required uplink resources in the UGS and rtPS algorithm (RU GS,RrtP S) are
RU GS= R1×
4
i=1
Ni. (1)
Fig. 6. Number of required uplink resources vs. number of users
RrtP S= R4× 3 i=1 Ni+ 4 i=1 Ri· Ni. (2)
Here, Ri is the number of required uplink resources for
transmitting voice packets generated by each data rate of EVRC (Li).
In case of the UGS-AD and Lee’s algorithms, although the operations of these algorithms are not the same, the numbers of required uplink resources (RU GS−AD,RLee) are exactly the
same. So, in this paper, we analyze and discuss only UGS-AD algorithm.RU GS−AD is given as
RU GS−AD = R1× 3 i=1 Ni+ R4· N4. = RLee. (3)
Also, the number of required uplink resource in our proposed algorithm (RP RD) can be represented by
RP RD =
4
i=1
Ri· Ni. (4)
When we use QPSK 1/2 for transmitting voice data packets,
R1, R2, R3, and R4 are 6, 4, 3, and 2 resource units. For performance analysis, we assume an OFDMA system, and one resource unit consists of 48 OFDM subcarriers. In addition, we assume that the size of generic MAC header is 6 bytes, and the RTP/UDP/IP headers are compressed by robust header compression (ROHC) [2], [9]. The sizes of uncompressed and compressed RTP/UDP/IP header are 40 bytes and 3 bytes. In case that NT ot = 20, N1, N2, N3, and N4 are 6, 1,
1, and 12, respectively. Hence, RU GS, RrtP S, RU GS−AD,
and RP RD are 120, 83, 72, and 67. We can show that our
proposed algorithm has the smallest the number of required uplink resources compared with the UGS, rtPS, UGS-AD, Lee’s algorithms. Fig. 6 shows the number of required uplink resources against the number of voice users.
In case that the voice users use another voice codec that has only on-off transitions, by equations 3 and 4, we can show
Fig. 7. Maximum number of users vs. MCS levels
that the numbers of required uplink resources in the UGS-AD (Lee’s) and our proposed algorithms are nearly the same.
B. VoIP Capacity by Analysis of Resource Utilization
By using RU GS, RrtP S, RU GS−AD, and RP RD, we
can calculate the maximum supportable number of voice users in each algorithm (NM U GS,NM rtP S,NM U GS−AD,
NM P RD). With the total number of uplink resources (RT ot
= 140), NM U GS=TTmfvc ×RRUGST ot,NM rtP S =TTmfvc ×RRrtP ST ot , NM U GS−AD = TTmfvc × RT ot RUGS−AD, and NM P RD = Tvc Tmf × RT ot
RP RD. In these equations, Tmf is a MAC frame duration
in IEEE 802.16 systems, 5ms, and Tvc is a frame duration
of EVRC, 20ms. In case of QPSK 1/2, NM U GS,NM rtP S,
NM U GS−AD, andNM P RD are 93, 131, 149, and 164. We
can also show that our proposed algorithm can support the largest number of voice users compared with the UGS, rtPS, UGS-AD, Lee’s algorithms. Fig. 7 shows the maximum num-ber of voice users (VoIP capacity) against several Modulation and Coding Scheme (MCS) levels.
C. Throughput
With the sizes of generic MAC header (LM H) and
com-pressed RTP/UDP/IP header (LU H), we can calculate
down-link throughput of each algorithm (SU GS,SrtP S,SU GS−AD,
SLee,SP RD). Sop = Tvc Tmf × 4 i=1 {(LM H+ LU H+ Li) × Ni} s.t.4 i=1 Ni< NM op. (5)
In the UGS, rtPS, UGS-AD, Lee’s, and proposed algorithms, ‘op’ in equation 5 is the same as ‘UGS’, ‘rtPS’, ‘UGS-AD’, ‘Lee’, and ‘PRD’.
D. Packet Transmission Delay
Except the rtPS algorithm, the packet transmission delays of the UGS, UGS-AD, Lee’s, and our proposed algorithms (Ttx U GS, Ttx U GS−AD, Ttx Lee, Ttx P RD) are represented
as
Ttx op= Tdsf+ Tttg+ Tq op. (6)
Tdsf andTttgare the durations of the downlink subframe and
the transmit transition gap (TTG) in IEEE 802.16 systems.
Tdsf and Tttg are constants and Tq op is queuing delay of
each algorithm.
Also, by using the durations of the uplink subframe Tusf
and the receive transition gap (RTG) (Trtg), and queuing delay
of the rtPS algorithm (Tq rtP S), the packet transmission delay
of the rtPS algorithm (Ttx rtP S) can be given as
Ttx rtP S = 3 i=1 pi× (Tdsf+ Tusf+ Tttg+ Trtg) + Tdsf+ Tttg+ Tq rtP S = Tmf × 3 i=1 pi+ Tdsf+ Tttg+ Tq rtP S. (7)
In equation 7, similar toTdsf andTttg,Tusf andTrtg are also
constants. And,Tdsf+Tusf+Tttg+Trtg= Tmf = 5ms. Since
the rtPS algorithm experiences the polling process in full, half and quarter rates, an addition delay would be generated in this algorithm. In this analysis, we assume that the polling process causes the delay of one MAC frame duration. However, in case that many voice users are activated, the polling process can cause the delay of more MAC frame durations. Although, the rtPS algorithm has additional delay terms compared with any other algorithm, we must not conclude that the rtPS algorithm has the largest packet transmission delay among these algorithms, because a dominant factor of packet trans-mission delay is a queuing delay of each algorithm. Tq op
is mainly determined by RT ot
Rop andNT ot. The relationship of
these elements is as follows.
Tq op ∝RT ot
Rop
, NT ot. (8)
In Section V, through simulation results, we prove that these relationships are correct.
V. SIMULATION RESULTS
For simulation results, we assume the IEEE 802.16 OFDMA system. A MAC frame consists of 36 symbols (time domain) and 1024 subcarriers (frequency domain). In this simulation, voice packets are sent by using QPSK 1/2. So, one basic resource unit is the same as six bytes. Also, we assume that durations of full, half, quarter and eighth rates are290, 40, 70 and600ms, respectively. Other simulation parameters are the same as we assume in Section IV.
Fig. 8 shows the total throughputs of the conventional and proposed algorithms. Through this figure, we can show that the values of throughput saturation points for the UGS, rtPS, UGS-AD (Lee’s) and our proposed algorithms are 93, 131, 149 and 164, respectively, and the throughput of our proposed
Fig. 8. Total throughput vs. number of users
algorithm is much larger than that of any other algorithms. The value of saturation point in each algorithm is the same with the maximum number of voice users in each algorithm in Fig. 7 The saturation of the throughput in each algorithm is generated due to resource limitation. In case of our proposed algorithm, because the number of required uplink resources is the smallest among these algorithms, the throughput is saturated later compared with any other algorithm. In addition, since we did not assume aggregations of voice packets in each data rate, the UGS algorithm has the worst performance in this simulation. However, if we assume the aggregations of voice packets, the throughputs of the UGS, UGS-AD, and Lee’s algorithms could be somewhat increased.
As shown in Fig. 9, we can show that the values of restricted points for the UGS, rtPS, UGS-AD (Lee’s) and our proposed algorithms are 101, 142, 162 and 176, respectively, and the values of the restricted points for packet transmission delay are larger than that of total throughput of each algorithm. This is because delay bound for packet transmissions is not0 ms, but 60 ms. In this simulation, in the consideration of maximum end-to-end delay bound (285 ms by ITU-T), backbone delay, packet processing delay, and handset playback buffer delay, we assume that delay bound of packet transmission is60ms [10]. Similar to the performance of throughput for each algorithm, in case of our proposed algorithm, the maximum supportable number of users is the largest in IEEE 802.16 systems. We can show that our proposed algorithm can support more 74%, 24%, and 9% voice users compared with the UGS, rtPS, and UGS-AD (Lee’s) algorithms, respectively.
VI. CONCLUSIONS
In this paper, we have proposed a novel uplink scheduling algorithm (ertPS) for VoIP services with variable data rates and silence suppression in IEEE 802.16 systems. In spite of the fact that there are some scheduling algorithms for the VoIP services, such as the UGS, rtPS, UGS-AD, and Lee’s algorithm, these algorithms have some problems of the waste of uplink resources, MAC overhead and additional access
Fig. 9. Packet transmission delay vs. number of users
delay owing to polling process. Especially, in case of the UGS-AD and Lee’s algorithms, although these algorithms partially can solve the problems of the UGS and rtPS algorithms, they are not perfect solutions for supporting VoIP services with variable data rates and silence suppression. However, our proposed algorithm can perfectly solve these problems of the conventional algorithms. Through the performance analysis and simulation results of resource utilization, VoIP capacity, throughput, and packet transmission delay, we have shown that our proposed algorithm has the best performance among these algorithms in IEEE 802.16 systems. In particular, our proposed algorithm can support more 74%, 24%, and 9% voice users compared with the UGS, rtPS, and UGS-AD (Lee’s) algorithms, respectively. We can say that our proposed algorithm is the best algorithm for VoIP services with variable data rates and silence suppression in IEEE 802.16 systems.
REFERENCES
[1] IEEE 802.16-2004, ”IEEE Standard for Local and Metropolitan Area Networks — Part 16: Air Interface for Fixed Broadband Wireless Access Systems,” Jun. 24, 2004.
[2] IEEE 802.16e/D10-2005, “IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems - Amendment for Physical and Medium Access Control Layers for Combined Fixed and Mobile Operation in Licensed Bands,” Aug., 2005.
[3] Data-Over-Cable Service Interface Spec., DOCSIS 2.0, “Radio Frequency Interface Spec. - CM-SP-RFIv2.0-I06-040804,” Aug. 4, 2004
[4] IEEE C802.16e-04/503 “UGS with Activity Detection for 802.16e,” Nov. 16, 2004.
[5] Howon Lee, T. Kwon, D. Cho, “An Enhanced Uplink Scheduling Al-gorithm Based on Voice Activity for VoIP Services in IEEE 802.16d/e System,” IEEE Communications Letters, pp. 216-218, Aug. 2005. [6] TIA/EIA/IS-127, “Enhanced variable rate codec, speech service option 3
for wideband spread spectrum digital systems,” 1996.
[7] IEEE C802.16e-04/522r3 “Extended rtPS for VoIP services,” Nov. 16, 2004.
[8] IEEE C802.16e-04/523r1 “Header compression specific Convergence Sublayer,” Nov. 16, 2004.
[9] C. Bormann et al., “RObust Header Compression(ROHC): Framework and Four Profiles: RTP, UDP, ESP, and Uncompressed,” IETF Network Working Group, RFC 3095, Jul. 2001.
[10] ITU-T Recommendation G.114, “One-way transmission time,” May, 2003.