Реферат: Evaluating the GPRS Radio Interface for Different Quality of Service Profiles
Evaluating the GPRS Radio Interfacefor Different Quality of Service Profiles
Abstract. This paper presents a discrete-eventsimulator for the General Packet Radio Service (GPRS) on the IP level. GPRS isa standard on packet data in GSM systems that will become commerciallyavailable by the end of this year. The simulator focuses on the communicationover the radio interface, because it is one of the central aspects of GPRS. Westudy the correlation of GSM andGPRS users by astatic and dynamic channel allocation scheme. In contrast to previous work, ourapproach represents the mobility of users through arrival rates of new GSM andGPRS users as well as handover rates of GSM and GPRS users from neighboringcells. Furthermore, we consider users with different QoSprofiles modeled by a weighted fair queueing scheme.The simulator considers a cell clustercomprising seven hexagonal cells. We provide curves for average carried trafficand packet loss probabilities for differentchannelallocation schemes and packet priorities as well as curves for averagethroughput per GPRS user. A detailed comparison between static and dynamicchannel allocation schemes is provided.
1Introduction
The General Packet Radio Service(GPRS) is a standard from the European Telecommunications StandardsInstitute (ETSI) on packet data in GSM systems [6], [14]. By adding GPRSfunctionality to the existing GSM network, operators can givetheirsubscribers resource-efficient wireless access to external Internetprotocol-bases networks, such as the Internet and corporate intranets. Thebasic idea of GPRS is to provide a packet-switched bearer service in a GSMnetwork. As impressively demonstrated by the Internet, packet-switched networksmake more efficient use of the resources for burstydata applications and provide more flexibility in general. In previous work,several analytical models have been developed to study data services in a GSMnetwork. Ajmone Marsan etal. studied multimedia services in a GSM network by providing more than one channelfor data services [1]. Boucherie and Litjens developed an analytical model based on Markov chainanalysis to study the performance of GPRS under a given GSM call characteristic[4]. For analytical tractability, they assumed exponentially distributedarrival times for packets and exponential packet transfer times, respectively.On the other hand, discrete-event simulation based studies of GPRS wereconducted. Meyer et al. focused on the performance of TCP over GPRS underseveral carrier to interference conditions and coding schemes of data [10].Furthermore, they provided a detailed implementation of the GPRS protocol stack[11]. Malomsoky et al. developed a simulation basedGPRS network dimensioning tool [9]. Stuckmann et al. studied the correlation of GSM and GPRSusers with the simulator GPRSim [13]. This paperdescribes a discrete-event simulator for GPRS on the IP level. The simulator isdeveloped using the simulation package CSIM [12] and considers a cellcluster comprising of seven hexagonal cells. Thepresented performance studies were conducted for the innermost cell of theseven cell cluster. The simulator focuses on the communication over the radiointerface, because this is one of the central aspects of GPRS. In fact, the airinterface mainly determines the performance of GPRS. We studied the correlationof GSM and GPRS users by a static and dynamic channel allocation scheme. Afirst approach of modeling dynamic channel allocation was introduced by Bianchiet al. and is known as Dynamic Channel Stealing (DCS) [3].
The basic DCS concept is totemporarily assign the traffic channels dedicated to circuit-switched connections but unusedbecause statistical traffic fluctuations. This can be done at no expense interms of radio resource, and with no impact on circuitswitchedservices performance if the channel allocation to packet-switched services is
permitted only for idle trafficchannels, and the stolen channels are immediately released when requested bythe circuit-switched service. In contrast to the models developed in [4], [9],[10], and [11], our approach additionally represents the mobility of usersthrough arrival rates of new GSM and GPRS users as well as handover rates ofGSM and GPRS users from neighboring cells. Furthermore, we consider users withdifferent QoS profiles modeled by a weighted fair queueing scheme according to [5]. The remainder of thepaper is organized as follows. Section 2 describes the basic GPRS networkarchitecture, the radio interface, and different QoSprofiles, which will be considered in the simulator. In Section 3 we describethe software architecture of the GPRS simulator, details about the mobility ofGSM and GPRS users, the way we modeled quality of service profiles, and theworkload model we used. Results of the simulation studies are presented inSection 4. We provide curves for average carried traffic and packet lossprobabilities for different channel allocation schemes and packet priorities aswell as curves for average throughput per GPRS user.
3 The SimulationModel
We consider a cluster comprising ofsever hexadiagonal cells in an integrated GSM/GPRSnetwork, serving circuit-switched voice and packet-switched data calls. Theperformance studies presented in Section 4 were conducted for the innermost cellof the seven cell cluster. We assume that GSM and GPRS calls arrive in eachcell according to two mutually independent Poisson processes, with arrivalrates ëGSM and ëGPRS, respectively. GSM calls are handled circuit-switched, so that onephysical channel is exclusively dedicated to the corresponding mobile station.After the arrival of a GPRS call, a GPRS session begins. During thistime a GPRS user allocates no physical channel exclusively. Instead the radiointerface is scheduled among different GPRS users by the Base StationController (BSC). Every GPRS user receives packets according to a specifiedworkload model. The amount of time that a mobile station with an ongoing callremains within the area covered by the same BSC is called dwell time. Ifthe call is still active after the dwell time, a handover toward an adjacentcell takes place. The call duration is defined as the amount of timethat the call will be active, assuming it completes without being forced toterminate due to handover
failure. We assume the dwell time tobe an exponentially distributed random variable with mean 1/µh,GSMfor GSM calls and 1/µh,GPRSfor GPRS calls. The call durationsare
also exponentially distributed withmean values 1/µGSMand 1/µGPRSfor GSM and
GPRS calls, respectively. To exactlymodel the user behavior in the seven cell cluster, we have to consider thehandover flow of GSM and GPRS users from adjacent cells. At the boundary cellsof the seven cell cluster, the intensity of the incoming handover flow cannotbe
specified in advance. This is due tothe handover rate out of a cell depends on the
number of active customers withinthe cell. On the other hand, the handover rate into
the cell depends on the number ofcustomers in the neighboring cells. Thus, the
iterative procedure introduced in[2] is used to balance the incoming and outgoing
handover rates, assuming that theincoming handover rate ëh GSM
in i ,
( ) ( ) −1 computed at step i-1.
Since in the end-to-end path, thewireless link is typically the bottleneck, and given
the anticipated traffic asymmetry,the simulator focuses on resource contention in the
downlink (i.e., the path BSC →BTS→MS) of the radio interface. Because of theanticipated traffic asymmetry the amount of uplink traffic, e.g. induced by
acknowledgments, is assumed to benegligible. In the study we focus on the radio
interface. The functionality of theGPRS core network is not included. The arrival
stream of packets is modeled at theIP layer. Let N be the number of physical channels available in the cell. Weevaluate the performance of two types of radio resource sharing schemes, whichspecify how the cell capacity is shared by GSM and GPRS users:
the static scheme;that is the cell capacity of N physical channels is split into
NGPRS channels reserved for GPRSdata transfer and NGSM = N — NGPRS channels
reserved for GSM circuit-switchedconnections.
the dynamic scheme;that is the N physical channels are shared by GSM and
GPRS services, with priority for GSMcalls; given n voice calls, the remaining
N-n channels are fairly shared byall packets in transfer.
In both schemes, the PDCHs are fairly shared by all packets in transfer up to a
maximum of 8 PDCHsper IP packet («multislot mode») and a maximum of 8 packets
per PDCH [6].
The software architecture of thesimulator follows the network architecture of the
GPRS Network [14]. To accuratelymodel the communication over the radio
interface, we include thefunctionality of a BSC and a BTS. IP packets that arrive at
the BSC are logically organized intwo distinct queues. The transfer queue can hold
up to Q n = ⋅8packets that are served according to a processor sharing service
discipline, with n the number ofphysical channels that are potentially available for
data transfer, i.e. n = NGPRS underthe static scheme and n = N under the dynamic
scheme. The processor sharingservice discipline fairly shares the available channel
capacity over the packets in thetransfer queue. An arriving IP packet that cannot enter
the transfer queue immediately isheld in a first-come first-served (in case of one
priority) access queue that canstore up to K packets. The access queue models the
BSC buffer in the GPRS network. Upontermination of a packet transfer, the IP
packet at the head of the accessqueue is polled into the transfer queue, where it
immediately shares in the assignmentof available PDCHs. For this study, we fix the
modulation and coding scheme to CS-2[14]. It allows a data transfer rate of 13,4
kbit/sec on one PDCH. Figure 1 depictsthe software architecture of the simulator.
Figure 1.Software Architecture of GSM/GPRS Simulator
To model the different quality ofservice profiles GPRS provides, the simulator
implemented a Weighted Fair Queueing (WFQ) strategy. The WFQ scheduling
algorithm can easily be adopted toprovide multiple data service classes by assigning
each traffic source a weightdetermined by its class. The weight controls the amount
of traffic a source may deliverrelative to other active sources during some period of
time. From the schedulingalgorithm's point of view, a source is considered to be
active if it has data queued at theBSC. For an active packet transfer with weight wi
the portion of the bandwidth Âi(t) allocated at time t to thistransfer should be
( ) ( ) = ⋅∑
where the sum over all active packettransfers at time t. The overall bandwidth at time
t is denoted by B(t) which isindependent of t in the static channel allocation scheme.
The workload model used in the GPRSsimulator is a Markov-modulated Poisson
Process (MMPP) [7]. It is used to generate the IPtraffic for each individual user in
the system. The MMPP has beenextensively used for modeling arrival processes,
because it qualitatively models thetime-varying arrival rate and captures some of the
important correlations between the interarrival times. It is shown to be an accurate
model for Internet traffic whichusually shows self-similarity among different time
scales. For our purpose the MMPP isparameterized by the two-state continuous-time
Markov chain with infinitesimalgenerator matrix Q and ratematrix Ë:
The two states represent bursty mode and non-bursty mode of the arrival process.
The process resides in bursty mode for a mean time of 1/áand in non-bursty modefor
a mean time of 1/ârespectively. Such anMMPP is characterized by the average
arrival rate of packets, ëavgand the degree of burstiness, B. The former is given by:
1 2
The degree of burstinessis computed by the ratio between the burstyarrival rate and
the average arrival rate, i.e., B = ë1/ëavg.
4Simulation Results
Table 1 summarizes the parametersettings underlying the performance experiments.
We group the parameters into threeclasses: network model, mobility model, and
traffic model. The overall number ofphysical channels in a cell is set to N = 20
among which at least one channel isreserved for GPRS. The overall number of GPRS
users that can be managed by a cellis set to M = 20. As base value, we assume that
5% of the arriving calls correspondto GPRS users and the remaining 95% are GSM
calls. GSM call duration is set to120 seconds and call dwell time to 60 seconds, so
that users make 1-2 handovers onaverage. For GPRS sessions the average session
duration is set to 5 minutes and thedwell time is 120 seconds. Thus, we assume
longer “online times” and slowermovement of GPRS users than for GSM users. The
average arrival rate of data is setto 6 Kbit/sec per GPRS user corresponding to 0.73
IP packets per second of size 1 Kbyte.
Parameter
Figure 2 presents curves for carrieddata traffic and packet loss probabilities due to
buffer overflow in the BSC for thestatic channel allocation scheme and one packet
priority. For GPRS 1, 2, and 4 PDCHs are reserved, respectively. The remaining
channels can be used by GSM calls.With 4 PDCHs the system overloads at an arrival
rate of 0.8 GSM/GPRS users persecond. This corresponds to an average of 12 GPRS
users in the cell (see Figure 7). InFigure 3 we present corresponding curves for the
dynamic channel allocation scheme.For GPRS 1, 2, and 4 PDCHs are reserved,
respectively but more PDCHs can be reserved «on demand». That meansthat
additional PDCHscan be reserved if they are not used for GSM voice service. From
Figure 3 we observe that for lowtraffic in the considered cell GPRS makes
effectively use of the on demand PDCHs. For example if 1 PDCH is reserved GPRS
utilizes up to 2 PDCHsat an arrival rate of 0.4 GSM/GPRS users per second. But
with increasing load the overallperformance of GPRS decreases because of
concurrency among GPRS users, andmore important, priority of GSM users over the
radio interface. In comparison withthe static channel allocation scheme we conclude
that the combination of reserved PDCHs and on demand PDCH leads to a better
utilization of the scarce radiofrequencies. The only advantage of the static channel
allocation scheme is that it can berealized more easily.
Figure 4 presents a comparison ofoverall channel utilization and average
throughput per GPRS user for thestatic and dynamic channel allocation scheme. For
the static scheme we reserved 2 and4 PDCHs respectively and for the dynamic
scheme only 1 PDCH. We observe ahigher overall utilization of physical channels by
the dynamic scheme. Comparing thedynamic with the static scheme for 2 PDCHs we
detect a slightly higher throughputfor low traffic load for dynamic channel allocation.
This results from the high radiochannel capacity available to GPRS users in this case.
They can utilize up to 8 PDCHs for their transfer (in contrast to 2 PDCHs in the static
scheme). When load increases, GSMcalls allocate most of the physical channels.
Thus, throughput for GPRS usersdecreases very fast. In the static scheme (4 PDCHs)
the decrease in throughput is not sofast, because GSM calls do not effect the PDCHs.
In an additional experiment, westudy the performance loss in the GSM voice
service due to the introduction ofGPRS. Figure 5 plots the carried voice traffic and
voice blocking probability fordifferent numbers of reserved PDCHs. The results are
valid for both channel allocationschemes because of the priority of GSM voice
service over GPRS. The presentedcurves indicate that the decrease in channel
capacity and, thus, the increase ofthe blocking probability of the GSM voice service
is negligible compared to thebenefit of reserving additional PDCHs for GPRS users.
Figure 6 shows carried data trafficand packet loss probabilities for the dynamic
channel allocation scheme anddifferent packet priorities. For GPRS1 PDCH is
reserved. Weights for packets withpriority 1 (high), 2 (medium), and 3 (low) and
percentages of GPRS users utilizingthese priorities are given in Table 1. We observe
that for low traffic in theconsidered cell most channels are covered by packets of low
priority. This is due to the highportion of low priority packets (60%) among all
packets sharing the radio interface.With increasing load medium priority packets and
at last high priority packetssuppress packets of lower priority and therefore the
utilization of PDCHsfor low and medium priority packets decreases. For a call arrival
rate of up to 2 calls per second theloss probability of high priority packets is still less
than 10-5 and therefore the correspondingcurve is omitted in Figure 6.
Figure 7 presents curves for averagenumber of GPRS users in the cell and
blocking probabilities of GPRSsession requests due to reaching the limit of M active
GPRS sessions. We observe that for2% GPRS users the maximum number of 20
active GPRS sessions is not reached.Therefore, the blocking probability remains very
low. For 10% GPRS users andincreasing call arrival rate, the average number of
sessions approaches its maximum.Thus, some GPRS users will be rejected. It is
important to note that the curves ofFigure 7 can be utilized for determining the
average number of GPRS users in thecell for a given call arrival rate. In fact, together
with the curves of Figure 2 and 3,we can provide estimates for the maximum number
of GPRS users that can be managed bythe cell without degradation of quality of
service. For example, for 5% GPRSusers and 1 PDCHs reserved, in the static
allocation scheme a packet lossprobability of 10-3 can be guarantied until the call
arrival rate exceeds 0.4 calls persecond, i.e., until there are on the average 6 active
GPRS users in the cell. For thedynamic allocation scheme a packet loss probability of
10-3 can be guarantied until thecall arrival rate exceeds 0.6 calls per second
corresponding to 9 active GPRS usersin the cell on average. Figure 8 investigates the impact of the maximum numberof GPRS user per cell to the performance of GPRS for the dynamic channelallocation scheme with 1 PDCH reserved. Of course, the expected number of GPRSusers should be less than the maximum number in order to avoid the rejection ofnew GPRS sessions. On the other hand, the maximum number of active GPRSsessions must be limited for guaranteeing quality of service for every activeGPRS session even under high traffic. The tradeoff between increasingperformance for allowing more active GPRS sessions and the
increasing blocking probability forGPRS users is illustrated by the curves of Figure 8.
Conclusions
This paper presented adiscrete-event simulator on the IP level for the General Packet Radio Service(GPRS). With the simulator, we provided a comprehensive performance study ofthe radio resource sharing by circuit switched GSM connections and packetswitched GPRS sessions under a static and a dynamic channel allocation
scheme. In the dynamic scheme weassumed a reserved number of physical channels permanently allocated to GPRSand the remaining channels to be on-demand channels that can be used by GSMvoice service and GPRS packets. In the static scheme no ondemandchannels exist. We investigated the impact of the number of packet data
channels reserved for GPRS users onthe performance of the cellular network. Furthermore, three different QoS profiles modeled by a weighted fair queueingscheme were considered. Comparing both channel allocation schemes, we concludedthat the dynamic scheme is preferable at all. The only advantage of the staticscheme lies in its easy implementation. Next, we studied the impact ofintroducing GPRS on GSM voice service and observed that the decrease in channelcapacity for GSM is negligible compared to the benefit of reserving additionalpacket data channels for GPRS. With the curves presented we provide estimatesfor the maximum number of GPRS users that can be managed by the cell withoutdegradation of quality of service. Such results give valuable hints for networkdesigners on how many packet data channels should be allocated for GPRS and howmany GPRS session should be allowed for a given amount of traffic in order to guaranteeappropriate quality of service.