Joint Reliability of Medium Access Control and Radio Link ...

Joint Reliability of Medium Access Control andRadio Link Protocol in 3G CDMA Systems

Mainak Chatterjee, Giridhar D. Mandyam, Senior Member, IEEE, and Sajal K. Das, Member, IEEE

Abstract—In this paper, we study the reliability as offered jointly by the medium access control (MAC) and the radio link protocol (RLP)

for third generation (3G) code division multiple access (CDMA) standards. The retransmission mechanism supported at the RLP layer

has a considerable amount of delay (80-100 ms) associated with it; hence, it may not be able to support applications with strict delay

requirements. On the other hand, if retransmissions are also performed at a lower layer, such as the MAC layer, then the performance

of CDMA systems can be improved because of the fast retransmissions at the MAC layer. We show how the performance of RLP

improves with respect to three metrics—mean delay, throughput, and RLP recovery—when a finite number of retransmissions is

allowed at the MAC layer. Synthetically generated Web traffic is used as the application, the objects of which are fragmented into

equal-sized RLP frames. We consider soft packet combining support at the receiver, which effectively lowers the frame error rate. We

also consider the possibility of misinterpretations of acknowledgments. Simulation experiments are conducted to verify the

performance as offered jointly by the MAC and the RLP, particularly for cdma2000 standard and wideband CDMA (WCDMA) systems.

The improvement in the TCP throughput is also evaluated.

Index Terms—Radio link protocol, MAC, retransmissions, packet combing, cdma2000, WCDMA.

�

1 INTRODUCTION

WITH the proliferation and everyday use of the WorldWide Web (WWW) applications, it has become

desirable that such applications also be supported overwireless access networks. Following the tremendous suc-cess of wireless voice services, service providers havestarted offering a variety of wireless data services such asaudio, video streaming, file, and Web downloading. Thesewireless data services promise to offer significant revenuegrowth opportunity for service providers all over the world.

To bring WWW traffic to wireless mobile devices, it isimportant that a suitable protocol or standard be chosen tocater to the growing demands of data services over wirelesschannels such that a wide variety of multimedia traffic withdifferent quality of service (QoS) requirements can besupported. No matter which wireless data technology orradio interface dominates future wireless networks, the factremains that it will most likely rely on the Internet Protocol(IP) which is currently the most dominant internetworkingprotocol. As for the transport layer, the transmission controlprotocol (TCP) is still the major suite for IP and providesreliable end-to-end transmission in wireline networks [9] inwhich the channel error rates are extremely low. However,when TCP is used over a wireless network with high bit

error rates (BER), its performance severely degrades [8].Any packet loss at the wireless link is interpreted ascongestion by TCP, which then responds to it by reducingthe transmission window size, initiating the congestioncontrol mechanism, and resetting the retransmission time[16]. This misinterpretation of channel related losses andconsequent invocation of the congestion control mechanismby TCP causes an unnecessary reduction in the TCPthroughput. Several schemes have been proposed toalleviate the effects of noncongestion related losses overwireless links—both for wireless local area networks (see[19] and references therein) and cellular networks [8], [15],[6], [23], [27], [28]. The radio link protocol (RLP) is one suchscheme that has been used for cellular mobile networks forsupporting data traffic [8].

1.1 Radio Link Protocols (RLP)

The RLP is generally employed within the data link layer,i.e., between the physical layer and the TCP layer to helpshield the effect of the loss over wireless links from theTCP layer [8], [15]. As shown in Fig. 1, the RLP fragmentsan upper layer packet (a TCP segment in this case) intoseveral equal-sized RLP frames before transmitting over thewireless channel. (The last frame can be padded if needed.)A physical layer header is added to the RLP frame before itis transmitted at the physical layer. The fragmentation isdone to increase the granularity of the transmission. Inother words, in case of any error, an RLP frame which is of asmaller size is affected rather than the entire TCP segment.The RLP uses an Automatic Repeat reQuest (ARQ) errorrecovery mechanism to retrieve a lost RLP frame. Therecovery process is initiated by the receiver by requesting aretransmission of only the missing or erroneous frame. Forexample, let us consider that a TCP segment is fragmentedinto L number of RLP frames which are transmitted oneafter another. If, somehow, the ith frame, 1 � i � L, fails to

1584 IEEE TRANSACTIONS ON COMPUTERS, VOL. 54, NO. 12, DECEMBER 2005

. M. Chatterjee is with the Department of Electrical and ComputerEngineering, University of Central Florida, PO Box 162450 Orlando, FL32816-2450. E-mail: [email protected].

. G.D. Mandyam is with Nokia Research Center, 12278 Scripps SummitDrive, San Diego, CA 92131. E-mail: [email protected].

. S.K. Das is with the Center for Research in Wireless Mobility andNetworking (CReWMaN), Department of Computer Science and En-gineering, University of Texas at Arlington, PO Box 19015, Arlington, TX76019. E-mail: [email protected].

Manuscript received 8 July 2004; revised 20 June 2005; accepted 30 June 2005;published online 14 Oct. 2005.For information on obtaining reprints of this article, please send e-mail to:[email protected], and reference IEEECS Log Number TC-0232-0704.

0018-9340/05/$20.00 � 2005 IEEE Published by the IEEE Computer Society

reach the receiver, the receiver would request for a

retransmission of that frame. However, the receiver buffers

the other frames ð1; 2; � � � ; i� 1; iþ 1; � � �LÞ and, on recep-tion of the ith frame, reassembles the TCP segment and

delivers it to the upper layer. The recovery from theerroneous frames is done before the TCP timer expires so

that the TCP throughput remains unaffected.The RLP is usually sufficient to shield the physical layer

impairment from the TCP. However, RLP might still fail to

conceal losses due to two reasons: 1) Applications might

have strict delay requirements [33] and 2) the packet lossrate at the medium access layer (MAC) layer might be high.

If the packet loss rate is high, even for elastic data traffic, theRLP has no way to shield the physical impairment from the

TCP. If real-time traffic is considered, the application isoften assumed to be loss-tolerant. For example, if packets

are dropped in voice communication, human intelligence isable to find the consistency in speech. Similarly, in video

transmission, if packets or frames are lost, the decoder at

the receiver conceals the losses by interpolating past andfuture frames. However, if channel conditions are extremely

bad, error concealment becomes almost impossible. In suchcases, interpolation or interpretation of a signal may not

only be difficult, but also impossible. Therefore, it isrecommended that as many damaged packets as possible

are salvaged, as quickly as possible.

1.2 RLPs for CDMA Systems

The performance of RLPs for various code division multipleaccess (CDMA) systems have been studied over the years as

the standard evolved. Performance issues related to TCPand RLP interaction in the CDMA protocol stack have been

investigated in [8]. The impact of TCP source activity on the

call admission control for the cellular CDMA standard IS-95was studied in [33]. Support of data services over the IS-95

physical channels using RLP was proposed in [22]. For IS-99(the first IS-95 data standard), the performance evaluation

of TCP over RLP was shown in [21] and the performance forcircuit mode data services was shown in [15]. The

performance of TCP over RLP in the cdma2000 system [2]was shown in [25]. A negative acknowledgment-based

hybrid ARQ scheme was proposed in [36]. In [27], the

performance of TCP over 3G networks was evaluated in thepresence of correlated fading channels.

In all these standards and propositions, RLP has been theonly layer below TCP to shield the losses by triggeringretransmissions. Though this single layer of reliability fromRLP shields losses for TCP, there can still be someperformance limitations if the TCP applications have strictdelay requirements. To deal with such demanding applica-tionsor interactive services, it is necessary to incorporate a fastretransmission mechanism below the RLP. This was done byincorporating an ARQ mechanism at the MAC layer, thusproviding two layers of retransmission reliability. The benefitof MAC layer ARQ is that retransmissions can be done veryquickly, without notifying RLP at the upper layer.

1.3 Our Contributions

Our main motivation behind this research is to study thereliability offered jointly by the RLP and the MAC layers of3G CDMA standards, particularly for cdma2000 1X-EV-DV[2] and wideband CDMA (WCDMA) [1] systems insupporting data traffic. It can be noted that two-layerreliability was not supported in the previous CDMAstandards such as IS-99 [21] and cdmaOne [36]. In thispaper, we do not propose any additional reliability, but,instead, study the performance that can be expected by theincorporation of MAC layer retransmissions in 3G systems.(Preliminary versions of these results can be found in [10],[11].) More specifically:

. We analyze the performance of both cdma2000 andWCDMA systems and demonstrate how the incor-poration of MAC layer retransmissions can signifi-cantly improve the delay performance of the RLP.

. We define three performance metrics: mean delay,throughput, and the fraction of packets recovered bythe RLP for a different number of retransmissiontrials at the MAC layer. The performance analysis isdone assuming a fading channel and considering asoft packet combining mechanism at the receiver.

. We extend the analytical models to incorporatemisinterpretations of acknowledgments (though asmall fraction) at the MAC layer.

. We conduct simulation experiments to validate theanalytical model. It is observed that the delay dropsfrom 80-100 ms to less than 10 ms if the MAC layer isallowed at least one retransmission attempt, althoughthe decrease in delay is not substantial for morethan two retransmissions. The recovery rate frommissing frames at the RLP increases with anincreasing frame error rate and also when themisinterpretations of acknowledgments are high.

. After studying the improvement in RLP perfor-mance, we study the effect on the TCP throughputwith and without the two layers of retransmissions.

The rest of the paper is organized as follows: Section 2discusses the two-layer retransmission scheme supported bythe 3G CDMA systems. The performance analysis is pre-sented in Section 3, where mean delay and the fraction ofpackets recovered by the RLP are evaluated. Section 4presents the simulation model and results with respect tomean delay, throughput, and RLP recovery. In Section 5, we

CHATTERJEE ET AL.: JOINT RELIABILITY OF MEDIUM ACCESS CONTROL AND RADIO LINK PROTOCOL IN 3G CDMA SYSTEMS 1585

Fig. 1. Fragmentation of TCP segments into RLP frames.

show how misinterpretations of acknowledgments couldpossibly affect the performance of the cdma2000 andWCDMA systems. The improvement in the TCP throughputis shown in Section 6. Conclusions are drawn in the lastsection.

2 MAC RETRANSMISSIONS IN CDMA SYSTEMS

A big challenge in third generation (3G) CDMA systems ishow to handle a wide variety of multimedia services withdifferent QoS requirements. To meet these demands, mod-ifications andenhancementsarebeingmade to cdma2000andWCDMA standards. cdma2000 is being deployed in com-mercial networks to support voice and data on the same1.25 MHz carrier [4]. A higher data rate evolution, calledcdma2000 1X-EV-DV, has been standardized by the Tele-communications Industry Association (TIA) and has beenadded to IMT-2000 [3]. As specified by the Third GenerationPartnership Project (3GPP2) [2], cdma2000 1X-EV-DV pro-vides integrated voice with simultaneous high-speed packetdata services such as video, video-conferencing, and othermultimedia services at speeds of up to 4.8 Mbps. Further-more, cdma2000 1X-EV-DV is backward compatible withcdmaOne and cdma2000 1X, providing wireless operators aseamless network evolution.

On the other front, high-speed downlink packet access(HSDPA) [10] is one of the main propositions for WCDMAsystems. HSDPA allows instantaneous bit rates up to10 Mbps for best-effort packet data services with certainbounds on the delay and capacity. The HSDPA channel is aresource that is shared among several users in the mobilecommunication system. By using a fast scheduler located atthe base station, the HSDPA channel can be assigned to theuser transmitting with the current best channel (i.e., highestdata rate). The underlying idea is that only users with gooddownlink channel conditions will use the HSDPA channelsand other users will not be allowed to use these channels ifthey experience bad channel conditions. Some of the basicprinciples used in HSDPA are fast link adaptation, fastscheduling, and fast retransmissions of erroneously re-ceived packets.

2.1 Fast ARQ at the MAC

Let us now discuss how the RLP and MAC layers jointlyprovide the enhanced retransmission reliability. Fig. 2 showstheworkingof these two layers at the transmitter and receiverends. A TCP segment reaches the MAC layer after beingfragmented at the RLP. For every packet transmitted at theMAC layer, there is either anACK or aNACK. This enhancesthe response time for retransmissions since the RLP is aNACK-based protocol. The fast ARQmechanism (also calledMAC-ARQ) in HSDPA and cdma2000 1X-EV-DV is impor-tant to ensure that some performance loss can be recovered.There are two reasons why RLP cannot provide thefunctionality needed for MAC-ARQ:

1. In the process of selecting the base station with thestrongest signal for the cell selection, the RLPterminates at the last network element (e.g., basestation), resulting in network delays in servicingretransmission requests at the RLP layer. Fast cell

site selection (FCSS), a feature under considerationin HSDPA, provides mobility support for best effortdata services. One way to achieve this is by themobile terminal “echoing” information about pack-ets it has recently received over the shared channelduring a transition from one base station to anotherso as to provide fast synchronization of the packettransmission queues between the new and the oldbase stations.

2. The high-speed downlink shared channel in HSDPAand the forward packet data channel in cdma20001X-EV-DV, which carry payload for best effort datausers (in both time and code multiplexed modes)might contain several protocol data units (PDUs),not all of which come from the RLP. In fact, somePDUs might come directly from the layer above theRLP. As a result, the MAC-ARQ provides retrans-missions quickly in HSDPA, which employs a stop-and-wait hybrid ARQ method. In this method, eachpacket received by the receiver must be acknowl-edged on a dedicated feedback channel to thetransmitter. This dedicated feedback takes the formof the reverse acknowledgment indicator withvalues of þ1 for an ACK, or �1 for a NACK.However, the receiver does not discard the receivedsoft information associated with the incorrectlyreceived packet. Rather, it buffers the data andcoherently combines the buffered data with thereceived soft information of the retransmission of thebad packet [34]. This type of packet combiningprovides increased reliability in CDMA systems.

2.2 n-Phase ARQ

Both HSDPA and cdma2000 1X-EV-DV use an n-phase stop-

and-wait MAC-ARQ [18]. By “n-phase,” we mean that

multiple ARQ instances are employed in consecutive time

slots (e.g., 5-ms frame durations). For instance, let us assume

that three ARQ channels are used as shown in Fig. 3. Then, in

time slot t, the receiverwill receive a packet corresponding to

phase 1. In time slot tþ 1, the receiver will receive a packet

corresponding to phase 2. In time slot tþ 2, the receiver will

receive a packet corresponding to phase 3. Again, in time slot

tþ 3, the receiver will receive a packet corresponding to

phase 1, and so on. The receiver must keep separate packets


Fig. 2. Two layers of retransmission.

received from different phases for packet combining andpacket acknowledgments. However, once any packet for anyphase is received correctly, the receiver may deliver thepackets to the higher layers (e.g., RLP).

3 PERFORMANCE MODELING

Let us now calculate the mean delay and the fraction(percentage) of packets recovered by the RLP, withoutconsidering the time taken for a packet to get scheduled fortransmission at the RLP. We consider the delay suffered by apacket before being correctly received by the RLP at thereceiver and then delivered to the upper layer. We assumethat the transmission and propagation delays are negligible.That is, if a packet is successfully transmitted at the firsttrial, then the delay incurred is zero. This assumption canbe relaxed by adding an offset (with a statistical fluctuation)to the delay value. If N denotes the number of ARQ phasesand T the frame duration, it takes about N � T ms for theACK to reach the transmitter. Only then can the transmitterremove the packet from its transmission buffer. If a packetdoes not successfully reach the receiver at the first trial,then, on the receipt of a NACK, it will undergo retransmis-sion. A delay will be incurred while the MAC attempts itsfast ARQ mechanism to retransmit and recover the missingpacket. Let the system parameter MAXRETRANS denotethe maximum number of retransmissions allowed at theMAC layer. If the packet is not recovered even afterMAXRETRANS retransmissions by the MAC layer, the RLPtriggers its own retransmission mechanism.

By definition, zero delay is incurred by a packet whichdoes not undergo any retransmission at RLP or MAC. Afinite delay is incurred only when a packet fails and isconsequently retransmitted. The total delay (D) experiencedby a packet to successfully reach the receiver can potentiallyhave two components: 1) DMAC , the delay due to the MAC-ARQ mechanism and 2) DRLP , the delay due to the RLPretransmissions. If a packet is recovered by the MAC layer,then the RLP does not need to use its retransmission and,hence, DRLP ¼ 0. If the MAC layer fails to recover a packet,then RLP invokes its retransmission and, hence, both thedelay components will be nonzero. Thus, the total delay is

D ¼ DMAC þDRLP : ð1Þ

In the following, we derive expressions for DMAC andDRLP . The expression for DRLP will be conditioned to the factthat the MAC layer failed even after the maximum numberof allowed retransmissions. Let us first briefly discuss theunderlying channel characteristics and the frame error rates(FERs) observed due to packet combining.

3.1 Channel Characteristics

In a wireless environment, the characteristics of the radiochannel are very important since the channel conditionsvary with time and space. There are several factors thataffect the link attenuation. The hindrance may be due tomultipath fading, shadowing, or any other noise source.The most commonly used additive white Gaussian noise(AWGN) channels do not represent the channel conditionsaccurately and, hence, a more realistic view of the channel isnecessary for any involved analysis or experimentation.Moreover, the channel model should include the correlatederrors, which is most often the case. Therefore, we considera 1-path Rayleigh fading model [26] for the losses, whichgenerally has the worst behavior among multipath Rayleighmodels due to its inherent lack of diversity [38].

3.2 Packet Combining Mechanism

In packet data services, when the receiver detects anerroneous packet in error, it requests a retransmission.However, the receiver does not have to discard the softinformation associated with the incorrectly received packet.Rather, it can buffer the data and coherently combine it withthe received soft information on the retransmission of thesame packet [34]. This is because there is a 3-db gain in theEb=N0 (Eb is the energy per bit and N0 is the spectral noisedensity) value if the retransmitted packet is soft-combinedwith the stored packet under quasistatic channel conditions[34]. This is because the retransmitted packet will doublethe power of the received signal; hence, the gain is 10 log 2 �3 db. If the packet is not correctly decoded even after thecombination, it is retransmitted for the second time (ifMAXRETRANS � 2). With the second retransmitted packet,the gain in Eb=No would be 6-db if two copies of the samepacket are retransmitted. That is, with every trial ofretransmission, there is approximately a gain of 3-db. Thistype of packet combining mechanism provides increasedreliability in CDMA systems. In order to ensure that thereceiver does not try to combine packets from one ARQ


Fig. 3. An example of 3-phase ARQ.

phase with another, it is assumed that the outbandsignaling is sent on a forward control channel concurrentlywith each frame that the receiver receives on the forwardshared channel.

3.3 Effective FER

We consider 1-path Rayleigh channel at 3 Km/h with 12 rate

8-PSK (phase shift key) modulation in the personalcommunication system PCS 1800-1900 MHz frequency.According to the simulation results reported in [29], theeffective FER falls to approximately one-third of its valuefor every transmission in a cdma2000 1X-EV-DV system.The FER values for successive retransmissions as obtainedin [29] are shown in Table 1. Although the reduction in theeffective FER could not be generalized, it works well withRayleigh fading channels. For example, if the FER is 0.3 forthe initial transmission, then, on successive retransmissions,the effective FER would be 0.095, 0.031, 0.0095, and so on,decreasing by approximately one-third every time. Thecorresponding FERs for WCDMA as obtained from [20] isshown in Table 2, for which the decrease in successive FERsis approximately a little less than half. In general, we canassume that there is a drop in the FER by a factor of C forevery retransmission, where C � 0:33 for cdma2000 1X-EV-DV (Table 1) and C � 0:45 for WCDMA systems (Table 2).

3.4 Analysis of Delay and RLP Recovery

Letusnowderive expressions formeandelayand the fractionof packets recoveredby theRLPwhenMAXRETRANS ¼ M.Assume that p is the rawFER offered by the physical channel.When a packet is transmitted, the MAC usually waits for acertain number of ARQ phases (say, N) for the ACK. Let thethroughput be defined as the ratio of the number of successfulpackets received to the total number of packets undergoingtransmission, including retransmissions.TheRLP recovery,R,is defined as the fraction of packets recovered by the RLPwhen theMAC fails to deliver a packet correctly.Wewill firstderive the expressions for the total delay (D) and RLPrecovery (R) for one transmission (M ¼ 1) and two transmis-sions (M ¼ 2) and then generalize them. Since the frame errorrate is finite, the MAC layer will fail with certain probabilityand, hence, the average case delay analysis will have aDRLP

component.Case I (M ¼ 1). The waiting time for a packet unsuccess-

fully received on the first transmission is N � T , where N is

the number of ARQ phases and T is the frame duration.

Therefore, the mean waiting time for all the packets at the

MAC layer is DMAC ¼ pNT , where p is the FER of the

wireless channel. The packets which will still fail to reach

the receiver successfully will be recovered by the RLP. The

packets being recovered by RLP would have a delay of M �NT ¼ NT since M ¼ 1. This is essentially the RLP delay for

triggering a retransmission. If M ¼ 0, the fraction of packets

recovered by RLP is p. Since M ¼ 1, the fraction is given by

Cp� p, assuming the FER value drops by a factor of C (as

noted earlier), which is different for cdma2000 and

WCDMA systems. Thus, the mean delay at the RLP is

DRLP ¼ Cp2 �NT . The total mean delay is given by

D ¼ DMAC þDRLP ¼ pNT þ Cp2NT: ð2Þ

The RLP recovery (R) measured by the fraction of packets

recovered by RLP is given by

R ¼ Cp2: ð3Þ

The factor C appears due to packet combining.Case II (M ¼ 2). Since the ACKs of ðp� 100Þ% of the

retransmitted packets are yet to arrive, these packets need

to be retransmitted again (second MAC retransmission)

and the effective FER observed so far will be Cp.

Therefore, the mean delay associated with the second

retransmission is Cp� pNT . Thus, the mean delay at the

MAC is DMAC ¼ pNT þ Cp� pNT . Since the fraction of

packets taken care of by RLP is Cp2 when M ¼ 1, the

fraction for M ¼ 2 would be Cp� Cp2 and the delay at the

RLP would be DRLP ¼ ðCpÞ2p� 2NT . Hence, the total

mean delay is obtained as

D ¼ pNT þ Cp� pNT þ ðCpÞ2p� 2NT: ð4Þ

The RLP recovery in this case would be

R ¼ ðCpÞ2p: ð5Þ

Case III (M > 2). Following in a similar manner, we

calculate mean delay at the MAC for any arbitrary M

retransmissions as


TABLE 1Effective FER Values for cdma2000 1X-EV-DV

TABLE 2Effective FER Values for WCDMA

DMAC ¼XMi¼1

ðCpÞi�1pNT

¼1Cp � ðCpÞM

1� CppNT:

ð6Þ

The delay at the RLP is

DRLP ¼ ðCpÞMp�MNT: ð7Þ

Therefore, the total delay is given by

D ¼1Cp � ðCpÞM

1� CppNT þ ðCpÞMp�MNT: ð8Þ

Additionally, the RLP recovery is given by

R ¼ ðCpÞMp: ð9Þ

3.5 Bound on M

Though the number of retransmissions at the MAC layer isa design parameter, it is primarily dictated by the RLPtimer. The RLP starts a timer when it pushes a packet to thelower (MAC) layer. If a packet fails to reach the receiver bythe expiration of the timer, the RLP invokes its ownretransmission mechanism to recover the missing packet.Therefore, we calculate the maximum number of retrans-missions that the MAC layer can afford before the RLPtimer expires and initiates its recovery process. In otherwords, the delay at the MAC layer must be less than theRLP timer, Trlp. We seek the maximum number ofretransmissions allowed at the MAC layer before the RLPintervenes. Therefore, the condition to be satisfied is

pNT þ ðCpÞ1pNT þ ðCpÞ2pNT þ � � � þ ðCpÞMpNT � Trlp;

ð10Þ

where M is the number of retransmission trials at the MAC.Since we are seeking the maximum number of retransmis-sions that the MAC can afford, we must find the maximum

value of M, Mmax, for which the inequality is satisfied.Therefore, Mmax is given by

Mmax ¼ maxfm : pNT þ ðCpÞ1pNT þ ðCpÞ2pNT þ � � �þ ðCpÞmpNT � Trlpg

¼ max m :1� ðCpÞmþ1

1� CppNT � Trlp

( )

¼ max m : ðCpÞmþ1 � 1� Trlpð1� CpÞpNT

� �

¼ max m : m � logCp 1� Trlpð1� CpÞpNT

� �� 1:

�ð11Þ

SinceMmax must be an integer, we take the floor functionsince the ceiling function will violate (10). Therefore, Mmax

is given by

Mmax ¼ logCp 1� Trlpð1� CpÞpNT

� �� 1

� �: ð12Þ

Taking the floor function also ensures that the time takenfor Mmax retransmissions at the MAC layer does not trigger

the RLP timer. Trlp is a system parameter which isnegotiated at the time of data connection establishment. Alarger value of Trlp gives more scope for MAC to recoverpackets. However, it might delay delivery of segments tothe upper layer (e.g., TCP), thus affecting the delayrequirements of the data application. So, depending onthe application requirement and the corresponding TCPtimeout value, the RLP timer is chosen.

4 SIMULATION STUDY

To validate the analytical model, we develop a UNIX-basedsimulator, perform extensive simulation experiments, andmeasure the average delay and the fraction of packetsrecovered by the RLP for the cdma2000 system. Thoughmultiple RLP frames can be mapped to a physical layerframe by the proper choice of codes from the code-tree [35],we assumed that one fixed-size RLP frame is mapped to aphysical layer frame. The hyper text transport protocol(HTTP) was used for the data application. We chose such amodel because of its interactive parameters, i.e., theparameters that also reflect the end-user response. ThoughHTTP is considered a non-real-time application, the down-load time is important to the end user and, hence, somedelay bounds are needed.

4.1 HTTP Traffic Model

Instead of investigating the nature of HTTP traffic, wesynthetically generate such traffic by using the resultsobtained in [12]. The basic model of HTTP is shown in Fig. 4inwhich a packet call represents the download of aWeb pagerequested by a user. It usually has a main page followed bysome embedded objects. A new request (packet call) isimmediately generated after the expiration of the viewingperiod. Themodel is similar to anON/OFF source,where theONstate represents the activity of a page request and theOFFstate represents a silent period after all objects in that page areretrieved. The download time of a page follows Weibulldistribution [12], the mean of which depends on the under-lyingbandwidth of thewireless channel.Wehave considereda data rate of 76.8 Kbps (9,600 bytes/s). Each object (mainpage and embedded objects) of the HTTP traffic is fragmen-ted into multiple equal-sized RLP packets. Other statisticsand parameters used to generate the HTTP traffic are showninTable 3. Theway eachRLPpacket is quickly acknowledgedis discussed next.


Fig. 4. Web page traffic scenario.

4.2 Fast ARQ at the MAC

We emphasize that the analysis is flexible enough to handlenewer advanced packet data access methods for bothupstream and downstream traffic. Though the analysisdoes not specifically follow 3GPP and 3GPP2 assumptions,the analysis can be readily adapted to new physical layersetups that use RLP and hybrid-ARQ. For example, theauthors in [17] describe the latest 1X-EV-DV evolution(enhanced reverse link) with frame durations as low as10 ms. Though the frame duration can vary depending onthe modulation and coding schemes, the largest frameduration is set to 20 ms.

We consider a configuration of the cdma2000 1X-EV-DVsystem where the transmitter transmits one RLP packet ineach 5 ms physical layer frame and waits for the ACK. If theACK does not arrive in 20 ms (equal to four ARQ phases,which is specific to cdma2000) [14], then the frame isretransmitted immediately. It can be seen from Fig. 5 thatframes with sequence numbers 0; 1; 2; 3; � � � are beingtransmitted. Frames 0 and 2 are undergoing retransmissionbecause of nonreceipt of ACKs. The ACK timers of thesepackets are again reinitialized. In a practical scenario, suchas the downlink, it is not necessary that the receiver deliverthe ACK/NACK precisely at the slot boundaries, each ofwhich is 5 ms. The actual physical layer boundary is moreprecise. This is due to the fact that, normally, when usingcoherent receivers in the reverse link, the base stationsuffers from some processing delay. The allowed number ofretransmission trials is varied between 1 and 3. If a packet isnot successfully received or combined at the receiver evenafter the maximum number of MAC retransmissions, thenthe RLP retransmission is triggered. Other parameters usedfor the simulation are listed in Table 4.

4.3 Analytical and Simulation Results

Fig. 6 shows the total delay (D) for various MAXRE-TRANS values. Not much difference is observed betweenMAXRETRANS = 2 and MAXRETRANS = 3. This is dueto the fact that, after two transmissions, most of the

packets are recovered and the third retransmission ishardly required. Fig. 7 shows the throughput of thesystem which is the combined throughput due to MACand RLP. It can be seen that the throughput decreases asthe FER increases. This is obvious because larger FERdamages more packets during transmission and, thus, thenumber of retransmitted packets is also high. Fig. 8presents the efficiency with which the RLP recovers themissing packets. If the MAC-ARQ mechanism is turnedoff, effectively implying MAXRETRANS = 0, all themissing packets are recovered by the RLP. WhenMAXRETRANS = 1, there is a considerable drop in theRLP recovery because the MAC does most of the recoverywith just one retransmission. However, there is not muchimprovement if the maximum number of such retransmis-sions at the MAC layer is more than 1. With MAXRETRANS= 2 or 3, the RLP recovery is almost zero because virtuallyall the packets are recovered with three retransmissions.

5 MISINTERPRETATION OF ACK/NACK

So far, we have not considered the probability that the ACKor NACK packets might get corrupted. In reality, there is asmall probability that some of the ACKs and NACKs fromthe MAC-ARQ will be misinterpreted [22], [39]. Thedecoder at the transmitter might interpret a NACK as anACK and therefore would not transmit the packet, assum-ing correct reception. The RLP at the receiver will detect themissing packet and trigger its own retransmission mechan-ism to recover the packet. On the other hand, if an ACK isdecoded as a NACK, then a retransmission will be triggeredby the MAC if the number of retransmissions has notreached the maximum limit (i.e., MAXRETRANS). If thenumber of allowable retransmissions at the MAC isexhausted and a packet has not been ACKed, then theRLP will initiate the recovery process. No matter whichlayer does the recovery, it results in unnecessary (duplicate)retransmissions. It can be noted that any kind of wronginterpretation at the MAC can only be detected by the RLPand, thus, the reliability of the RLP cannot be ignored. Thisis where the real benefit and importance of RLP ismanifested. The percentage of error recovery by the RLPdepends on the percentage of misinterpretations, which wecall the “falseACK.”

5.1 Analytical Modifications

Due to the presence of the falseACKs, we need to modifythe analytical expressions for D and R derived inSection 3.4. Let f denote the percentage of falseACKs. The


TABLE 3Statistics for HTTP

Fig. 5. Number of ARQ phases (N) = 4 timing diagram.

TABLE 4Simulation Parameters

fraction of packets not reaching the RLP at the first trial is p

simply because p is the FER and one packet gets mapped to

one physical layer frame. Out of these, a fraction f will be

misinterpreted as ACK and the fraction ð1� fÞ will under-

go retransmission at the MAC layer, leading to an FER of

Cp. The fraction of packets not correctly reaching after the

first retransmission (if MAXRETRANS = 1) is pðCpÞð1� fÞ.Therefore, for M ¼ 1, due to the misinterpretations, the RLP

recovers an additional fraction of packets given by

AddM¼1 ¼ fpþ pðCpÞð1� fÞ: ð13Þ

If M ¼ 2, the fraction of packets misinterpreted on

the second retransmission is fpðCpÞð1� fÞ. The fraction

of packets not delivered correctly by the MAC is

pðCpÞ2ð1� fÞ2. Therefore, the additional fraction of packets

recovered by the RLP in this case is

AddM¼2 ¼ fpþ fpðCpÞð1� fÞ þ pðCpÞ2ð1� fÞ2: ð14Þ

In general, the additional fraction of packets recovered

by RLP is derived as

AddM ¼XM�1

i¼0

fpðCpÞið1� fÞi þ pðCpÞMð1� fÞM: ð15Þ

The summation term corresponds to the misinterpretations

and the second term corresponds to the fraction of packets

failing to reach the receiver even after MAXRETRANS

retransmissions. Therefore, the RLP recovery expression in

(9) is modified as


Fig. 6. Mean delay.

Fig. 7. Throughput.

R ¼ ðCpÞMpþAddM: ð16Þ

Clearly, the number of packets recovered by RLP

increases with misinterpretations. The delay expression

obtained in (8) will modify to

D ¼XMi¼1

ðCpÞi�1pNT þ ðAddMÞ �MNT: ð17Þ

Note that, if f ¼ 0, (16) and (17) will reduce to (9) and (8),

respectively.

5.2 Results with falseACK

We conducted simulation experiments by incorporating a

certain percentage of ACK misinterpretations. This

percentage is very small and is usually around 2 percent[39], so we varied f from 1 percent to 5 percent. Fig. 9 showsthat the mean delay due to simulation as well as analysis(from (8)) for MAXRETRANS = 2. We observe that the delayis more as compared to Fig. 6, where f ¼ 0. Fig. 10 showsthe percentage of RLP recovery for MAXRETRANS = 2 andthis percentage is more than that for MAXRETRANS = 1.This is because almost all the packets are recovered withtwo retransmissions at the MAC layer and the RLP mostlyrecovers the ones that were misinterpreted. It is expectedthat the packet recovery by RLP will be more at largervalues of FER and higher percentage of ACK misinterpreta-tions. However, we did not consider FER greater than30 percent because, at that high FER, there will be other


Fig. 8. RLP recovery.

Fig. 9. Mean delay (MAXRETRANS = 2).

severe consequences with maintaining the link. Fig. 11shows the degradation in the system performance in termsof throughput when M ¼ 2.

Comparing the simulation and analytical results, weobserve that they are in good agreement for FER < 0:2.However, theydiffer by less than 1

2 percent for FER> 0:2. Onepossible reason might be the assumption on the FER forsuccessive retransmissions. The FERs used for simulation(see Table 1) do not strictly fall by 1

3 for cdma2000 systems,which was assumed for obtaining the analytical results.

5.3 WCDMA Performance

Simulation experiments were also conducted for WCDMAsystems, using the same parameters as those used forcdma2000 simulations, except that the WCDMA frameduration was 2 ms (3 � 1.67 ms). The percentage of

misinterpretations was varied from f ¼ 1% to f ¼ 5%.

Fig. 12 depicts the mean delay for MAXRETRANS = 2.

Fig. 13 shows the degradation in the system performance in

terms of the throughput for MAXRETRANS = 2. The

percentage of RLP recovery for MAXRETRANS = 2 is

shown in Fig. 14. The RLP recovers more packets with

MAXRETRANS = 1 than with MAXRETRANS = 2. This

happens because, as explained earlier, almost all the packets

are recovered with two retransmissions at the MAC layer.

The RLP mostly recovers the ones which were misinter-

preted. It can be expected that the RLP recovery will be

more for even larger FER values. The value of C ¼ 0:45 for

WCDMA (see Table 2) can be used for analytical purpose.To compare the performance of cdma2000 1X-EV-DV and

WCDMA systems, we summarize in Table 5 the mean delay,


Fig. 10. RLP recovery (MAXRETRANS = 2).

Fig. 11. Throughput (MAXRETRANS = 2).

RLP recovery, and throughput of these two standards for

MAXRETRANS = 2 and misinterpretation fraction f ¼ 5%.

6 TCP THROUGHPUT ENHANCEMENT

With the two layers of retransmission, we evaluate the

throughput improvement at the TCP layer. Recall from

Section 1.1 that TCP segments are fragmented into multiple

RLP frames. If we assume that a TCP segment of size T bytes

is fragmented into equal-sized RLP frames, then the

number of such frames is L ¼ dTRe, where R is the payload

of each RLP frame. The actual size of the RLP frame would

be R plus some header information. For a TCP segment to

be reassembled successfully, all the L frames must be

received correctly. If one or more RLP frames fail, the TCP

segment is lost. Thus, the TCP segment loss probability,TCPloss, is given by

TCPloss ¼ 1� ð1� pÞL; ð18Þ

where p is the frame error probability at the physical layer.If, however, we assume an underlying RLP (1, 2, 3) inoperation, i.e., one retransmission at the first trial, tworetransmissions at the second, and three retransmissions atthe third trial, then the effective frame loss probability at theRLP layer is p7. Hence, with the RLP layer, the TCP segmentloss probability is modified to

TCPloss ¼ 1� ð1� p7ÞL: ð19Þ

Furthermore, if we assume that the MAC layer is also inoperation, allowing only one retransmission attempt (i.e.,


Fig. 12. Mean delay (MAXRETRANS = 2).

Fig. 13. Throughput (MAXRETRANS = 2).

MAXRETRANS = 1) to recover any lost or corrupted RLPframe, then the effective frame loss probability at the RLPlayer is ðp7Þ2 and (19) is modified to

TCPloss ¼ 1� ð1� p14ÞL: ð20Þ

We will assume that the TCP throughput is given by [30]:

STCP ¼MSS

RTT

ffiffiffiffiffiffiffiffiffiffiffiffi2bTCPloss

3

pþT0min 1;3

ffiffiffiffiffiffiffiffiffiffiffiffi3bTCPloss

8

pÞTCPlossð1þ32TCP2

loss

� ; ð21Þ

where MSS is the maximum segment size, RTT is theround trip time for the TCP ACKs, T0 is the TCPretransmission timer, and b is a system constant. T0 isevaluated as an exponentially moving average of theinstantaneous RTTs.

Substituting the value of TCPloss obtained from (18), (19),and (20), we can obtain the TCP throughput when 1) bothRLP and MAC are off, 2) RLP is on and MAC is off, and3) both RLP and MAC are on, respectively. Fig. 15 showsthe relative performance of these three cases. The RTT is

assumed to be 0.2 seconds, T0 ¼ 2�RTT , and b ¼ 2, as per

the traces obtained in [30]. MSS is assumed to be 1,460

bytes. RLP frames were assumed to be of constant size of

80 bytes. The transmit window size for the TCP is assumed

to grow without limit. The improvement in TCP throughput

is due to two reasons. First, the fragmentation of the TCP

segments into RLP frames prevents the entire TCP segment

from being retransmitted, if lost. Second, due to faster

retransmissions and soft combining, the effective frame loss

probability is drastically reduced.

7 CONCLUSIONS

As the demand for real-time wireless data services

increases, more efficient and faster protocols have to be

designed for timely delivery of data packets. In this paper,

we demonstrated how cdma2000 1X-EV-DV and WCDMA

promise to be such protocols which can deliver real-time

data over the wireless medium. This is made possible with

the help of the joint reliability offered by both the MAC and


Fig. 14. RLC Recovery (MAXRETRANS = 2).

TABLE 5Performance of cdma2000 1X-EV-DV and WCDMA

RLP layers. The MAC layer retransmissions can recover

missing or damaged packets very quickly without notifying

the upper layer (e.g., RLP) and thus reduce the delay from

about 100 ms to less than 10 ms. Due to the soft packet

combining mechanism, the effective FER experienced by a

packet is lowered for every successive retransmission. We

analyzed the system performance in terms of delay and the

fraction of packets recovered by the RLP for a finite number

of retransmissions at the MAC layer. The analysis is

corroborated with simulation experiments using syntheti-

cally generated HTTP traffic. We also considered misinter-

pretations of the ACKs at the MAC layer. Results

demonstrated that the fast retransmission mechanism at

the MAC layer enhances the overall TCP throughput.

ACKNOWLEDGMENTS

The authors are grateful to the anonymous referees for

insightful comments which greatly helped them improve

the quality of the paper.

REFERENCES

[1] http://www.3gpp.org, 2005.[2] http://www.3gpp2org, 2005.[3] http://www.itu.org, 2005.[4] TIA/EIA/IS-2000-2, “Physical Layer Standard for cdma2000

Spread Spectrum Systems,” Mar. 2000.[5] TIA/EIA/IS-95-B, “Mobile Station—Base Station Compatibility

Standard for Dual Mode Spread Spectrum Systems,” 1999.[6] TIA/EIA/IS-707-A.2, “Data Service Options for Spread Spectrum

Systems: Radio Link Protocol,” Mar. 1999.[7] A. Bakre and B.R. Badrinath, “I-TCP: Indirect TCP for Mobile

Hosts,” Proc. 15th Int’l Conf. Distributed Computing Systems, pp. 136-143, May 1995.

[8] G. Bao, “Performance Evaluation of TCP/RLP Protocol Stack overCDMA Wireless Links,” ACM Wireless Networks J., vol. 2, pp. 229-237, 1996.

[9] R. Caceres and L. Iftode, “Improving the Performance of ReliableTransport Protocols in Mobile Computing Environments,” IEEE J.Selected Areas in Comm., vol. 13, pp. 850-857, June 1995.

[10] M. Chatterjee, G.D. Mandyam, and S.K. Das, “Fast ARQ in HighSpeed Downlink Packet Access for WCDMA Systems,” Proc.European Wireless, pp 451-457, Feb. 2002.

[11] M. Chatterjee, S.K. Das, and G.D. Mandyam, “MAC LayerRetransmissions in 1XTREME,” Proc. 13th IEEE Int’l Symp.Personal, Indoor and Mobile Radio Comm. (PIMRC), vol. 3, pp. 1452-1456, 2002.

[12] H.K. Choi and J.O. Limb, “A Behavioral Model of Web Traffic,”Proc. Int’l Conf. Network Protocols (ICNP), pp. 327-334, 1999.

[13] S. Choi, Y. Choi, and I. Lee, “IEEE 802. 11 MAC-Level FEC withRetransmission Combining,” IEEE Trans. Wireless Comm., Jan.2005.

[14] Y.-J. Choi and S. Bahk, “Scheduling for VoIP Service in cdma20001x EV-DO,” Proc. IEEE Int’l Conf. Comm. (ICC) 2004, vol. 3,pp. 1495-1499, 2004.

[15] A. Chockalingam and G. Bao, “Performance of TCP/RLP ProtocolStack on Correlated Fading DS-CDMA Wireless Links,” IEEETrans. Vehicular Technology, vol. 49, no. 1, pp. 28-33, Jan. 2000.

[16] D.E. Comer, Internetworking with TCP/IP, vol. 1. Prentice-Hall,1991.

[17] R.T. Derryberry and Z. Pi, “Reverse High-Speed Packet DataPhysical Layer Enhancements in cdma2000 1xEV-DV,” IEEEComm. Magazine, pp. 41-47, Apr. 2005.

[18] M. Dottling, J. Michel, and B. Raaf, “Hybrid ARQ and AdaptiveModulation and Coding Schemes for High Speed DownlinkPacket Access,” Proc. 13th IEEE Int’l Symp. Personal, Indoor, andMobile Radio Comm. (PIMRC), vol. 3, pp. 1073-1077, 2002.

[19] H. Elaarag, “Improving TCP Performance over Mobile Net-works,” ACM Computing Surveys, vol. 34, no. 3, pp. 357-374, Sept.2002.

[20] F. Frederiksen and T. Kolding, “Performance Modeling ofWCDMA/HSDPA Transmission/HARQ Schemes,” Proc. IEEEVehicular Technology Conf. (VTC), pp. 472-476, Fall 2002.

[21] A.S. Joshi, M.N. Umesh, A. Kumar, T. Mukhopadhyay, K. Natesh,S. Sen, and A. Arunachalam, “Performance Evaluation of TCPover Radio Link Protocol in TIA/EIA/IS-99 Environment,” Proc.IEEE Int’l Conf. Personal Wireless Comm., pp. 216-220, 1999.

[22] J.M. Harris and M. Airy, “Analytical Model for Radio LinkProtocol for IS-95 CDMA Systems,” Proc. IEEE 51st VehicularTechnology Conf. (VTC), vol. 3, pp. 2434-2438, Spring 2000.

[23] C. Geßner, R. Kohn, J. Schniedenharn, and A. Sitte, “Layer 2 andLayer 3 of UTRA-TDD,” Proc. IEEE 51st Vehicular Technology Conf.(VTC), vol. 2, pp. 1181-1185, Spring 2000.

[24] H. Kaaranen, A. Ahtiainen, L. Laitinen, S.K. Naghian, and V.Niemi, UMTS Networks: Architecture, Mobility, and Services. Wiley& Sons, 2001.


Fig. 15. TCP throughput.

[25] F. Khan, S. Kumar, K. Medepalli, and S. Nanda, “TCPPerformance over cdma2000 RLP,” Proc. IEEE 51st VehicularTechnology Conf. (VTC), vol. 1, pp. 41-45, Spring 2000.

[26] W.C.Y. Lee, Mobile Cellular Telecommunications. McGraw-Hill,1995.

[27] H. Lin and S.K. Das, “Performance Study of Link Layer and MACLayer Protocols in Supporting TCP in 3G CDMA Systems,” IEEETrans. Mobile Computing, vol. 4, no. 5, pp. 489-501, Sept./Oct. 2005.

[28] H. Lin and S.K. Das, “Performance Study of CDMA Systems withRLP and MAC Layer Retransmissions,” Proc. IEEE Int’l Symp.Modeling, Analysis, and Simulation of Computer and Telecomm.Systems (MASCOTS), pp. 313-320, Oct. 2002.

[29] P. Luukkanen, Z. Rong, and L. Ma, “Performance of 1XTREMESystem for Mixed Voice and Data,” Proc. IEEE Int’l Conf. Comm.(ICC), vol. 5, pp. 1411-1415, 2001.

[30] J. Padhye, V. Firoiu, D. Towsley, and J. Kurose, “Modeling TCPThroughput: A Simple Model and Its Empirical Validation,” Proc.ACM SIGCOMM pp. 303-314, 1998.

[31] S. Parkvall, E. Dahlman, P. Frenger, P. Beming, and M. Persson,“The Evolution of WCDMA towards Higher Speed DownlinkPacket Data Access,” Proc. IEEE 53rd Vehicular Technology Conf.(VTC), vol. 3, pp. 2287-2291, 2001.

[32] M. van der Schaar, S. Krishnamachari, S. Choi, and X. Xu,“Adaptive Cross-Layer Protection Strategies for Robust ScalableVideo Transmission over 802. 11 WLANs,” IEEE J. Selected Areas inComm., vol. 21, no. 10, pp. 1752-1763, Dec. 2003.

[33] S.K. Sen, J. Jawanda, K. Basu, N.K. Kakani, and S.K. Das, “TCPSource Activity and Its Impact on Call Admission Control inCDMA Voice/Data Network,” Proc. Fourth ACM/IEEE Int’l Conf.Mobile Computing and Networking (MobiCom) 1998, pp. 276-283,1998.

[34] S. Souissi and S.B. Wicker, “A Diversity Combining DS/CDMASystem with Convolutional Encoding and Viterbi Decoding,”IEEE Trans. Vehicular Technology, vol. 44, no. 2, pp. 304-312, May1995.

[35] Y.C. Tseng and C.M. Chao, “Code Placement and ReplacementStrategies for Wideband CDMA OVSF Code Tree Management,”IEEE Trans. Mobile Computing, vol. 1, no. 4, pp. 293-302, Oct.-Dec.2002.

[36] M.N. Umesh, A.S. Joshi, A. Kumar, and T. Mukhopadhyay, “ANAK Based Hybrid Type II ARQ Scheme for cdmaOne/cdma2000Systems,” Proc. IEEE Vehicular Technology Conf. (VTC), vol. 5,pp. 2596-2600, Fall 1999.

[37] Y. Xiao, H. Li, and S. Choi, “Protection and Guarantee for Voiceand Video Traffic in IEEE 802.11e Wireless LANs,” Proc. IEEEINFOCOM, vol. 3, pp. 2152-2162, Mar. 2004.

[38] A.J. Viterbi, CDMA: Principles of Spread Spectrum Communication.Addison-Wesley, 1995.

[39] X. Zhaoji and B. Sebire, “Impact of ACK/NACK Signalling Errorson High Speed Uplink Packet Access (HSUPA),” Proc. IEEE Int’lConf. Comm. (ICC), May 2005.

Mainak Chatterjee received the PhD degreefrom the Department of Computer Science andEngineering at the University of Texas atArlington in 2002. Prior to that, he completedthe BSc degree in physics (Hons) from theUniversity of Calcutta in 1994 and the ME degreein electrical communication engineering from theIndian Institute of Science, Bangalore, in 1998.He is currently an assistant professor in theDepartment of Electrical and Computer Engi-

neering at the University of Central Florida. His research interestsinclude economic issues in wireless networks, applied game theory,resource management and quality-of-service provisioning, ad hoc andsensor networks, CDMA data networking, and link layer protocols. Heserves on the executive and technical program committee of severalinternational conferences.

Giridhar D. Mandyam received the BSEEdegree (magna cum laude) from SouthernMethodist University (SMU) in 1989, the MSEEdegree from the University of Southern Califor-nia (USC) in 1993, and the PhD degree from theUniversity of New Mexico in 1996. He is thedirector of the Radio Systems Group in theRadio Communications Laboratory of NokiaResearch Center and the head of the NokiaResearch Center, San Diego. At SMU, he was a

University Scholar and Hyer Society Scholar. He held positions atRockwell International, Qualcomm Inc., and Texas Instruments beforejoining Nokia Research Center (NRC) in Dallas in 1998. In 2002, hebecame the director of the Radio Systems Group at NRC’s RadioCommunications Laboratory. In 2004, he became the first head of thenewest Nokia Research Center division in San Diego, California. Whileat NRC, he continued his participation in CDMA standardization work,which he started while at Texas Instruments. He was a contributor to thedevelopment of the cdma2000 wireless standard. More recently, he hasled efforts in building experimental radios and participated in researchinto “Beyond 3G” technologies. He is the inventor or coinventor of 10issued US patents. He has also published more than 50 conference andjournal papers and four book chapters. In addition, he was a guest editorfor a special issue of the Eurasip Journal on Applied Signal Processingentitled “3G Wireless Communications and Beyond.” He is a coauthor ofthe book Third-Generation CDMA Systems for Enhanced Data Services(Academic Press, 2002). Dr. Mandyam is a senior member of the IEEE.

Sajal K. Das received the BTech degree in 1983from Calcutta University, the MS degree in 1984from the Indian Institute of Science, Bangalore,and the PhD degree in 1988 from the Universityof Central Florida, Orlando, all in computerscience. He is a professor of computer scienceand engineering and also the founding directorof the Center for Research in Wireless Mobilityand Networking (CReWMaN) at the University ofTexas at Arlington (UTA). His current research

interests include resource and mobility management in wireless andsensor networks, mobile and pervasive computing, wireless multimediaand QoS provisioning, mobile Internet protocols, distributed processing,and grid computing. He has published more than 350 research papers,directed numerous funded projects, and holds five US patents inwireless mobile networks. He received the Best Paper Award from ACMMobiCom ’99, ICOIN ’01, ACM MSWIM ’00, and ACM/IEEE PADS ’97.He was also a recipient of UTA’s Outstanding Faculty Research Awardin Computer Science (2001 and 2003), College of EngineeringExcellence in Research Award (2003), and University Award forDistinguished Record of Research (2005). He is the coauthor of thebook Smart Environments: Technology, Protocols and Applications(John Wiley, 2005). He is the editor-in-chief of the Pervasive and MobileComputing journal and serves on the editorial boards of five internationaljournals, including IEEE Transactions on Mobile Computing, IEEETransactions on Parallel and Distributed Systems, and ACM/SpringerWireless Networks. He has served as general chair of IEEE WoWMoM’05, IWDC ’04, IEEE PerCom ’04, CIT ’03, and IEEE MASCOTS ’02;general vice chair of IEEE PerCom ’03, ACM MobiCom ’00, and HiPC’00-01; program chair of IWDC ’02 and WoWMoM ’98-99; TPC vicechair of CIT ’05 and ICPADS ’02; and as TPC member of numerousIEEE and ACM conferences. He is the vice chair of two IEEE technicalcommittees (TCPP and TCCC) and a member of the advisory boards ofseveral cutting-edge companies. He is a member of the IEEE and theIEEE Computer Society.

. For more information on this or any other computing topic,please visit our Digital Library at www.computer.org/publications/dlib.


Joint Reliability of Medium Access Control and Radio Link ...

Technology

medium access control

call admission control

radio link protocol

radio systems group

cdma wireless links

ieee intl conf

supporting data traffic

frame error rate