Packet multiplexers with adversarial regulated traf®c
qSrini Rajagopal
a, Martin Reisslein
b,1,*, Keith W. Ross
caManugistics, 2115, East Jefferson Street, Rockville, MD 20852-4999, USA
bDepartment of Electrical Engineering, Arizona State University, Tempe, AZ 85287-7206, USA
cInstitut EureÂcom, BP 193-06904 Sophia-Antipolis, France Received 2 June 2000; revised 8 May 2001; accepted 29 May 2001
Abstract
We consider a ®nite-buffer packet multiplexer to which traf®c arrives from several independent sources. The traf®c from each of the sources isregulated, i.e. the amount of traf®c that can enter the multiplexer is constrained by known regulator constraints. The regulator constraints depend on the source and are more general than those resulting from cascaded leaky buckets. We assume that the traf®c is adversarial to the extent permitted by the regulators. For lossless multiplexing, we show that if the original multiplexer is lossless it is possible to allocate bandwidth and buffer to the sources so that the resulting segregated systems are lossless. For lossy multiplexing, we use our results for lossless multiplexing to estimate the loss probability of the multiplexer. Our estimate involves transforming the original system into two independent resource systems, and using adversarial sources for the two independent resources to obtain a bound on the loss probability of the transformed system. We show that the adversarial sources are not extremal on±off sources, even when the regulator consists of a peak rate controller in series with a leaky bucket. We explicitly characterize the form of the adversarial source for the transformed problem. We also provide numerical results for the case of the simple regulator.q2002 Elsevier Science B.V. All rights reserved.
Keywords: Buffer-bandwidth tradeoff curve; Call admission control; Leaky bucket; Quality-of-service guarantees; Resource allocation; Statistical multi- plexing; Worst-case sources
1. Introduction
Consider a ®nite-buffer packet multiplexer to which traf®c arrives from several independent sources. For the multiplexer to provide quality of service (QoS) guarantees, such as limits on packet loss probabilities, it must have some knowledge about the traf®c characteristics of the sources.
Because the reliability of statistical models of traf®c is ques- tionable for many source types, in recent years there have been several studies on the performance of packet-switched nodes that multiplex regulated traf®c, e.g. traf®c which conforms to known constraints imposed by leaky buckets.
These studies suppose that the traf®c from the sources is adversarial to the extent permitted by the regulators [1±
14,16,19±21,23±27]. Some of these studies assume that the multiplexer provides deterministic QoS guarantees (e.g. no packet loss) whereas others assume that multiplexer
provides less stringent probabilistic QoS guarantees (e.g. a limit on the packet loss probability).
In a recent paper, LoPresti et al. [16] examine a packet- switched node with regulated traf®c. Motivated by earlier work of Elwalid et al. [9], LoPresti et al. consider both deterministic QoS guarantees and probabilistic QoS guaran- tees. They assume that each source is regulated by asimple regulator, namely, a regulator that consists of a peak-rate controller in series with a leaky bucket. For deterministic QoS, LoPresti et al. show that if the multiplexer has suf®- cient link bandwidth and buffer capacity toprovide lossless multiplexing, then the multiplexer's buffer and bandwidth can be allocated among the sources so that the resulting segregated systems are lossless. For probabilistic QoS, they develop a new approach to estimate the loss probabil- ity. Speci®cally, they transform the two-resource (band- width and buffer) allocation problem into two independent single-resource allocation problems; they then analyze these simpler, independent resource problems, taking on±off periodic sources for their adversarial sources.
Although the simple regulator is a popular policing mechanism within several standards bodies, it has been observed that it is often a poor characterization of a source's
0140-3664/02/$ - see front matterq2002 Elsevier Science B.V. All rights reserved.
PII: S0140-3664(01)00359-0
www.elsevier.com/locate/comcom
qA shorter version of this paper has appeared in Proceedings of IEEE Infocom, San Francisco, CA, March 1998.
* Corresponding author. Tel.:11-480-965-8593; fax:11-480-965-8325.
E-mail address:reisslein@asu.edu (M. Reisslein).
1 www.eas.asu.edu/,mre.
worst-case traf®c. A tighter and more powerful characteri- zation is given by a more general regulator consisting of a cascade of multiple leaky buckets [10,15,25]. For example, when the sources are VBR video sources, it is often possible to admit signi®cantly more connections by replacing the simple regulator with cascaded-leaky-bucket regulators [25]. It is therefore desirable to extend the important work of LoPresti et al. [16] and Elwalid et al. [9] to the case of more general regulators.
In this paper, we reexamine the model of LoPresti et al.
[16] in the context ofgeneralized regulators, which are even more general than cascaded leaky buckets. We ®rst re- examine the lossless multiplexer of LoPresti et al., and extend their lossless results to generalized regulators.
Using elementary tools from calculus, we show that if the original multiplexer is lossless, then it is possible to allocate bandwidth and buffer to the sources so that the resulting segregated systems are also lossless. We determine the optimal resource allocations and show that the buffer-band- width tradeoff curve is convex for generalized regulators.
We also show that the segregation result does not necessa- rily hold for the delay-based QoS metric, even when the regulators are the simple regulators.
We then examine the multiplexer for probabilistic loss guarantees. We use our results for lossless multiplexing to estimate the loss probability of the multiplexer. As in Ref.
[16], our estimate involves the following three steps: (i) choose a point on the buffer-bandwidth tradeoff curve and transform the original system into two independent resource systems; (ii) use adversarial sources for the two independent resources to obtain a bound on the loss probability of the transformed system; and (iii) minimize the bound by search- ing over all points on the buffer-bandwidth tradeoff curve.
Our principle contribution for probabilistic loss guarantees is an explicit characterization of the adversarial source for the transformed problem in Step (ii). Importantly,the most adversarial source is not a periodic on±off source for the transformed problem consisting of two independent resources. In fact, even in the case of simple regulated sources as studied in Ref. [16], the most adversarial source is not a periodic on±off source. Thus, in addition to general- izing the theory in Ref. [16] to the case of general regulated sources, we provide the true adversarial source for the case of the simple regulator. We also provide an algorithm to calculate the estimate of the loss probability, assuming the truly adversarial sources.
We mention here that in Ref. [9] the original multiplexer problem is transformed into a bufferless multiplexer problem, and then the loss probability is bounded with the Chernoff bound. In this case, the worst-case adversarial sources are indeed on±off periodic sources. But when the original problem is transformed into a problem consisting of two independent resources, one bufferless resource and one buffered resource, the worst-case sources are no longer on±
off periodic sources, even for simple regulators.
This paper is organized as follows. In Section 2, we de®ne the model and the generalized regulators. In Section 3, we address lossless multiplexing. In Section 4, we address lossy multiplexing. In Section 5, we provide numerical results for lossy multiplexing of simple regulators, i.e. regulators consisting of a peak rate controller in series with a leaky bucket. We summarize our contributions in Section 6.
2. Regulated traf®c
We consider a link of rateCwhich is preceded by a ®nite buffer of capacityB. Let Jbe the number of sources that send traf®c to the buffer, and let j1;¼;J index the sources. Each sourcejhas an associated regulator function, denoted byej t;t$0:The regulator function constrains the amount of traf®c that thejth source can send over an time interval of lengtht toej t:More explicitly, if Aj t is the amount of traf®c that thejth source sends to the buffer over the interval0;t;thenAj ´is required to satisfy
Aj t1t2Aj t#ej t for allt$0; t$0:
Fig. 1 shows a multiplexer consisting of a link of rateC, a buffer of capacityB, andJsources with regulator functions, ej t;j1;¼J:
A popular regulator is the simple regulator, which consists of a peak-rate controller in series with a leaky bucket; for the simple regulator, the regulator function takes the following form:
ej t minnr1jt;sj21r2jto :
For a given source type, the bound on the traf®c provided by the simple regulator may be loose and lead to overly conservative admission control decisions. For many source types (e.g. for VBR video [25]), it is possible to get a tighter bound on the traf®c and dramatically increase the admission region. In particular, piece-wise linear regulator functions of the form
ej t minnr1jt;sj21r2jt;¼;sjLj1rLjjto
are easily implemented with cascaded leaky buckets [1,5]
and can lead to improved admission regions [25].
In this paper, we shall consider extremely general regu- lator functions, which include as special cases the forms mentioned above. To avoid certain trivialities, however, we shall always assume thatej 0 0;ej tis non-decreasing
Fig. 1. Link of capacityC, buffer of capacityB, andJregulators.
in t, and that ej t is subadditive in t (i.e. ej t11t2# ej t11ej t2 for all t1 and t2). Also, unless explicitly mentioned otherwise, we shall assume that each ej t is concave int. Let
e t XJ
j1
ej t
be the aggregate regulator function. Due to the concavity of the ej t0s; the aggregate regulator function e t is also concave.
Before preceding with our analysis of the lossless systems, it is convenient at this point to introduce some notation and state a few technical facts. Let e1j t denote the right derivative forej tande2j tdenote the left deri- vative forej t:Lete0j tdenote the derivative ofej twhen- ever the derivative exists att. Similarly de®nee1 t;e2 t
ande0 t:We will make use of the following fact: Ife tis differentiable attp, then all of theej t0s are differentiable at tp(due to the concavity of theej t0s).
3. Guaranteed lossless service and optimal segregation It is well known [4] that the amount of traf®c in the buffer does not exceedBmin, where
Bminmax
t$0 {e t2Ct}: 1
(To avoid trivialities we assume that the maximum is attained in Eq. (1).) Furthermore, due to sub-additivity, it is possible to de®ne traf®c functionsAj t;j1;¼;J;such that the buffer contents will attainBmin. Thus the minimum buffer size that will guarantee lossless operation is Bmin. Throughout the remainder of this section we assume that the multiplexer is lossless, i.e. we assume that the multi- plexer bufferBsatis®esB$Bmin:It will be useful to write Eq. (1) in a more convenient form. Ife tis differentiable then from Eq. (1) we have
Bmine tmax2Ctmax; 2
wheretmaxis any solution toe0 t C:More generally, there exists atmaxsuch that
e1 tmax#C#e2 tmax; 3
and anytmaxwhich satis®es Eq. (2) and also satis®es Eq. (3).
Throughout the remainder of this section, ®x a tmax that satis®es Eq. (3) (and therefore Eq. (2) as well).
We now address the following question: Is it possible to allocate bandwidth and buffer to theJsources so that each of the resulting segregated systems is also lossless? We shall see that the answer to this question is yes, but depends critically on the concavity of theej t0s:
To address this issue, consider a new system which consists of a link of ratecpreceded by a ®nite buffer. Suppose only the traf®c from source j is sent to this system. The
minimum buffer size that will ensure lossless operation is Bmin j;c max
t$0 {ej t2ct}: 4
We say that a collection ofJpositive numbersc1;¼;cJis a bandwidth allocation if c11¼1cJC: For a given bandwidth allocation, we create J segregated systems, with thejth segregated system having link ratecjand receiv- ing traf®c only from sourcej.
Theorem 1.
1. For all allocations we may have Bmin#PJ
j1Bmin j;cj:
2. If one or more of theej t0s is not concave then we may have Bmin #PJ
j1Bmin j;cjfor all allocations c1;¼;cJ: 3. If each ej t is concave then BminPJ
j1Bmin j;cpj where cpj e0j tmax if e t is differentiable at ttmax and where
cpj e1j tmax1ae2j tmax2e1j tmax
with
a C2e1 tmax e2 tmax2e1 tmax
ife tis non-differentiable at ttmax:
Proof. The proof of the ®rst claim follows from Eqs. (1) and (4):
Bmin max
t$0 {e t2Ct}max
t$0
XJ
j1
{ej t2cjt}
#XJ
j1
maxt$0 {ej t2cjt}XJ
j1
Bmin j;cj
For the second claim, we offer the following counter- example withJ2;C1:The envelope function for the
®rst source is:
e1 t
t if 0#t#1;
1 if 1#t#3;
11 t23 if 3#t#4;
2 ift$4:
8>
>>
>>
<
>>
>>
>:
The envelope function for the second source is:
e2 t 2t if 0#t#2;
4 ift$2:
(
It is easily seen that Bmin3 whereas Bmin 1;c11 Bmin 2;c2$10=3 for all allocations. Note that both e1 t
and e2 t are non-decreasing and sub-additive. However, e1 tis not concave.
For the third claim, we ®rst show thatcp1;¼;cpJis a feasi- ble allocation. Suppose thate tis differentiable attmax. Due to the concavity assumption, this implies that each of the ej t's is differentiable attmax. Thus
XJ
j1
cpj XJ
j1
e0j tmax e0 tmax C:
If e tis not differentiable at ttmax;then it is easy to show directly from the de®nition of the cpjs thatcp11¼1 cpJC:It remains to show thatBminPJ
j1Bmin j;cpj:For a ®xed transmission rate c, the concavity of theej t's and Eq. (4) imply
Bmin j;c ej tp2ctp; wheretpis anytthat satis®es
e1j t#c#e2j t: 5
By the de®nition ofcpj; e1j tmax#cpj #e2j tmax:
Thus,tmaxis atthat satis®es Eq. (5) forccpj:Therefore, Bmin j;cpj ej tmax2cpjtmax;
which in turn implies XJ
j1
Bmin j;cpj XJ
j1
ej tmax2cpjtmax e tmax2Ctmax
Bmin: B
From Theorem 1 we know that it is possible to allocate bandwidth and buffer so that the resulting segregated systems are lossless, provided that the regulator functions are concave. This result generalizes a result in Ref. [16], in which all regulators were assumed to be simple regulators.
This result also provides a motivation for the approach we take in Section 4 when we study probabilistic QoS.
Theorem 1 also gives fairly explicit formulas for these optimal allocations. In Section 3.1, we outline an ef®cient algorithm for calculating the allocations.
3.1. Algorithm to calculate allocations
In this section, suppose that each of the regulator func- tions takes the form of cascaded leaky buckets:
ej t min{r1jt;sj21r2jt;¼;sjLj1rLjjt}:
Without loss of generality, we may assume that
0sj1,sj2,¼,sjLj 6
and
r1j .r2j .¼.rLjj: 7
Let
Tjl sjl112sjl
rlj2rl11j ; l1;2;¼;Lj21:
In order to avoid trivialities, we assume that
Tj1 ,Tj2,¼,TjLj21: 8
With these assumptions, Tj1 ,Tj2,¼,TjLj21 are the breakpoints ofej t:
Here is an ef®cient algorithm for determining the optimal allocationscp1;¼;cpJde®ned in Theorem 1. First sortTj1;l 1;¼;Lj; j1;¼;J;in increasing order. Number them as T1;T2;¼;TL:These points are the break points ofe t:Letk be the maximumlsuch thate2 Tl#C:Note that to calcu- latee2 Tlit suf®ces to calculatee2j Tlforj1;¼;J;and to calculatee2j Tl;we can determine theljsuch thatTjlj # Tl,Tjlj11 and then set e2j Tl rljj11 if Tjlj,Tl and set e2j Tl rljj ifTjljTl:
Thetmax in Theorem 1 isTk. Once having determinedk,
®nd kj such that Tjkj #Tk,Tjkj11 and set cpj rkjj11 if Tjkj ,Tk or set cpj rkjj1a rkjj112rkjj if Tjkj Tk; whereais de®ned in Theorem 1 and can also be determined directly from therljs:
3.2. The buffer-bandwidth tradeoff curve
For a given link rateCletBmin Cbe the maximum buffer contents de®ned by Eq. (1).The functionBmin Cis usually referred to as thebuffer-bandwidth tradeoff curve[16] (it is also known as burstiness curve [17]). For a probabilistic analysis in the Section 4, it will be useful to understand the behavior of the buffer-bandwidth tradeoff curve. To this end, for each ®xed C let t C be a value oftmax that satis®es Eq. (3). It is easily seen thatt Cis non-increasing inC.
Theorem 2. Bmin Cis non-increasing and convex in C.
Proof. We ®rst show thatBmin Cis non-increasing. Let h.0:From Eq. (2) we have
Bmin C2Bmin C1h e t C2e t C1h
1t C1h C1h2t CC: 9
From the concavity ofe twe have e2 t C# e t C2e t C1h
t C2t C1h : 10
From Eq. (3) we have
c#e2 t C: 11
Combining Eqs. (9)±(11) gives
Bmin C2Bmin C1h$e2 t Ct C2t C1h
1t C1h C1h2t CC$Ct C2t C1h
1t C1h C1h2t CCt C1hh$0;
which proves the ®rst statement.
For the convexity ofBmin C;letC1#C2 and leth.0:
We must show
Bmin C21h2Bmin C2$Bmin C11h2Bmin C1: 12
By Eq. (2) it is equivalent to show
e t C21h2e t C21e t C12e t C11h
$t C21h C21h2t C2C22t C11h C11h
1t C1C1:
13
Using the arguments in the proof of monotonicity, we have
e t C22e t C21h
t C22t C21h #C21h 14
and
e t C12e t C11h
t C12t C11h $C1: 15
Combining Eqs. (13)±(15), we obtain Eq. (12). B From Theorem 2 we know that Bmin C is a decreasing convex function ofC. If each of the regulator functionsej t
is piecewise linear, then it is easily shown thatBmin Cis a decreasing convex piecewise-linear function. Using the arguments in the proof of Theorem 2, it is straightforward to show that the optimal allocationcpj for thejth segregated system is increasing inCand that the buffer requirement for thejth segregated system,Bmin j;cpj;is decreasing inC.
3.3. Delay metric
In Section 3.1, we showed how to allocate bandwidth so that Ð for lossless operation Ð the collective buffer requirement of the segregated system is equal to the buffer requirement of the multiplexed system. In other words, for the buffer metricwe can ®nd a bandwidth allocation such that the segregated system performs as well as the multi- plexed system. In this section, we brie¯y consider a natural delay metric. We show that it is not generally true that the segregated system performs as well as the multiplexed system for the delay metric.
For the multiplexed system, the maximum delay isd:
Bmin=C:For thejth segregated system with bandwidthcjthe
maximum delay isd j;cj:Bmin j;cj=cj:For a given allo- cation, we de®ne the maximum delay of the collective segregated system to be the maximum of the maximum delays of the individual segregated systems, that is, dseg: max
1#j#J d j;cj:
The following theorem draws comparisons between the maximum delay of the multiplexed system,d, and the maxi- mum delay of the collective segregated system,dseg.
Theorem 3.
1. For all allocations d#max1#j#Jd j;cj:
2. There exist concaveej t's such that d,max1#j#Jd j;cj for all allocations.
3. Ife1 t ¼eJ t(homogeneous regulator functions), then dmax1#j#Jd j; c=J:
Proof. From Theorem 1 we have
Bmin #XJ
j1
Bmin j;cj:
Dividing both sides of the above by Cc11¼1cJ and using the inequality
x11¼1xJ
y11¼1yJ # max
1#j#J
xj yj: we obtain
d# XJ
j1
Bmin j;cj XJ
j1
cj
# max
1#j#J
Bmin j;cj cj
( )
max
1#j#Jd j;cj;
which establishes the ®rst claim.
For the second claim, we offer the following example:
C1;J2;e1 t 10 for allt$0 and
e2 t 2t if 0#t#5;
10 if t$5:
(
From Eq. (1) we have d15; d 1;c1 10=c1; and d 2;c2 10=c225:It is easily seen that for all allocations max 10=c1;10=c225.15:
The third statement follows directly from Eq. (1) and the de®nitions ofdandd j;C=J: B
For the remainder of the paper, we will use the original buffer metric.
4. Statistical multiplexing with small loss probabilities For VBR sources the admission region can typically be made signi®cantly larger by allowing loss to occur with minute probabilities, e.g. loss probabilities on the order of 1027. In this section, we use our results of Section 3 to derive the worst-case loss probabilities for the multiplexer with regulated traf®c.
We consider the same system as de®ned in Section 2: the multiplexer consists of a link of rateCwhich is preceded by a ®nite buffer. There areJsources and thejth source has an associated regulator function, denoted byej t;t$0:In this section, we suppose that the system resources are not suf®- cient to provide guaranteed lossless service. In other words, we assume B,Bmin C; so that there exists arrival processes which meet the regulator constraints but cause the buffer to over¯ow. LetPlossdenote the expected fraction of time during which the buffer over¯ows. Our goal is to determine a bound forPlossthat holds forallcombinations of arrival processes which meet the regulator constraints. To this end, we follow the methodology in Ref. [16] (which in turn is inspired by the paper [9]).
Letaj tbe the rate at which sourcejtransmits traf®c at timet. We view {aj t;t$0} as a stochastic process. Our goal is to ®nd independent rate processes {aj t;t$0}j 1;¼;J;which maximize the loss probability over the class of all rate processes that meet the regulator constraints. To simplify the analysis, however, we only consider rate processes of the form
aj t bj t1uj;
where bj t is a deterministic periodic function with some periodTj, andujis a random variable, uniformly distributed over0;Tj:We assume that the phasesu1;¼;uJ are inde- pendent, which implies that the rate processes {aj t;t$0};
j1;¼;J; are also independent. We refer to bj t as a source-j rate function.
We say that a source-jrate functionbj tisfeasibleif Zt1t
t bj sds#ej t for allt$0; t$0: 16
Note that for a given rate function bj tand phase uj the amount of source-j traf®c sent to the multiplexer over the interval0;tis
Aj t Zt
0bj s1ujds:
Thus the regulator constraint
Aj t1t2Aj t#ej t for allt$0; t$0 is satis®ed if and only ifbj tis a feasible rate function.
As in Ref. [16], our derivation of a bound for Ploss
involves the following three steps: (i) choose a point on the buffer-bandwidth tradeoff curve and transform the origi- nal system into two independent resource systems; (ii) use adversarial rate functions for the two independent resources
to obtain a bound on the loss probability for the transformed system; (iii) minimize the bound by searching over all points on the buffer-bandwidth tradeoff curve. LoPresti et al. use an on±off rate function for their worst case rate function. Our approach differs from that of LoPresti et al.
[16] in two respects. First, we allow for generalized regula- tors as opposed to simple regulators. Second, we derive the true adversarial rate functions, and employ these true adver- sarial rate functions in the bound for Plossfor both simple and generalized regulators.
4.1. The virtual segregated system
Fix a point Cn;Bn on the buffer-bandwidth tradeoff curve, and consider a lossless multiplexer with total amount of bandwidthCn and buffer spaceBn. Because the system resource pair B;Clies below the buffer-bandwidth tradeoff curve, we must have eitherCn.CorBn.Bor both. For this lossless system, we use Theorem 1 to allocate band- widths cn1;¼;cnJ from Cn and buffers bn1;¼;bnJ from Bn
such that each of the corresponding Jsegregated systems is lossless. This collection ofJsegregated systems is called the virtual segregated system [16].
For each j1;¼;J; ®x a feasible rate function bj t:
Each rate function generates a stochastic arrival process
Aj t Zt
0bj s1ujds:
For this arrival process, letUjbe a random variable that corresponds to the steady-state utilization of thejth segre- gated system; similarly, letVjbe the random variable that corresponds to the steady-state buffer contents of the jth segregated system. Because theuj's are independent across the sources, U1;¼;UJ are independent of each other and V1;¼;VJ are independent of each other.
For these ®xed rate functions it can be argued [16] that
Ploss#Pn XJ
j1
Uj.C 0
@
1
A1Pn XJ
j1
Vj.B 0
@
1
A: 17
(The argument in Ref. [16] is for a simple regulator. It can be easily extended to our generalized regulators.) Pn PJ
j1Uj.Cis the probability that the sum of the band- width processesU1;¼;UJexceeds the nodal bandwidthC.
The subscriptn indicates that the steady-state random vari- ables U1;¼;UJ are computed based on the allocation of bandwidth Cn to the J segregated systems. Similarly, Pn PJ
j1Vj.Bis the probability that the sum of the buffer processes V1;¼;VJ exceeds the nodal buffer capacity B.
Again, the subscriptn indicates that the steady-state random variablesV1;¼;VJare computed according to the allocation of bufferBnto theJsegregated systems.
The Eq. (17) is the starting point of our own analysis.
Using the Chernoff bound we get
Ploss#min
a$0
YJ
j1
MnUj a eaC 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
; 1min
a$0
YJ
j1
MVnj a eaB 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
;
; 18
whereMUnj aandMVnj aare the moment generating func- tions ofUjandVj, respectively, i.e. MUnj a EeaUj and MVnj a EeaVj: Since Eq. (18) is valid for all points Cn;Bnon the buffer-bandwidth tradeoff curve, we have
Ploss# min
Cn;Bn min
a$0
YJ
j1
MUnj a eaC 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
; 1min
a$0
YJ
j1
MVnj a eaB 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
; 2
66 66 64
3 77 77 75:
19
We emphasize that the right-hand side of Eq. (19) depends on the ®xed feasible rate functions. In order to give a bound that holds for all feasible rate functions we need to maximize the right-hand side of Eq. (19) over the set of all feasible rate functions. To this end, we introduce the notion of asource-j adversarial rate function.
Corresponding to each choice of n;a; we say that a source-j rate function is adversarial if (i) it is feasible, and (ii) it has the largest value of MnUj a and MVnj a among all feasible source-j rate functions. Now suppose that we can ®nd the source-jadversarial rate functions for each choice of n;a;letUjp;Vjp;j1;¼;J;be the corre- sponding steady-state random variables. We then have the following bound onPloss:
Ploss# min
Cn;Bn min
a$0
YJ
j1
MUnjp a eaC 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
; 1min
a$0
YJ
j1
MVnjp a eaB 8>
>>
><
>>
>>
:
9>
>>
>=
>>
>>
; 2
66 66 64
3 77 77 75:
20
Note that by usingMUnpj aandMVnjp a;which corresponds to the source-jadversarial rate function, we have obtained in Eq. (20) a bound onPlossthat is valid for all combinations of feasible arrival functions. We now proceed to characterize the adversarial rate functions.
4.2. Adversarial sources
Throughout this section ®x an,aandj. We now focus on
determining a feasible rate function which maximizes both MnUj a andMVnj aover the set of feasible rate functions.
We assume that the regulator functions have the form ej t minnr1jt;sj21r2jt;¼;sjLj1rLjjto
:
Note thatej tis non-decreasing, concave, piecewise-linear and sub-additive. (The analysis that follows can easily be extended to the case of more general ej t which are non- decreasing, concave and sub-additive.) Without loss of generality, we also assume that Eqs. (6)±(8) hold. Note that the manner in which the allocations cn1;¼;cnJ are chosen (see Theorem 1) ensures thatrLjj#cnj #r1j for all j1;2;¼;J:
For a given feasible rate functionbj twith periodTj, the arrival rate at time t is aj t bj t1uj where uj is uniformly distributed over 0;Tj: Corresponding to this aj t arrival rate process, let vj t be the buffer contents and uj t be the link utilization at time t. Note that vj t
anduj tare periodic with periodTj. Also the steady-state random variables corresponding tovj tanduj thave distri- butions
P Vj#x 1 Tj
ZTj
0 1 vj s#xds and
P Uj#x 1 Tj
ZTj
0 1 uj s#xds:
Note that these distributions do not depend on the phaseuj
and are completely determined by the rate functionbj tand the link ratecnj:
Throughout the remainder of this section we treat the case cnj .rLjj:In Section 4.3, we deal with the simpler casecnj rLjj:Let
djmax t.0: ej t
t $cnj
: 21
Note that sincerLjj ,cnj #r1j and sinceej ´is an increasing concave function,djis a uniquely de®ned, ®nite and strictly positive number. We now de®ne an important class of rate functions. LetToffbe such that 0,Toff#djand let Tj ej Toff
rLjj :
Now consider a rate functionbj twith periodTjde®ned as follows:
bj t e1j t 0#t#Toff; 0 Toff #t#Tj: 8<
:
Such a rate function is pictured in Fig. 2.
This rate function is completely characterized by the parameter Toff. Note that the average arrival rate for this rate function is simply rLjj: Let Sj be the collection of all
Fig. 2. Example of a rate function in SetSj whenToff3 and ej t min{3t;2:510:5t}:
rate functions of this form. Each rate function inSjis iden- ti®ed through itsToffparameter.
We will show that the setSjhas the following important properties:
1. Each member ofSjis a feasible source-jrate function.
2. All members in Sj have identical MUnj a; and the members ofSjmaximizeMnUj aover the set of all feasi- ble source-jrate functions.
3. The member inSjwhich has the largest MnVj ahas, in fact, the largestMVnj aamong all feasible source-j rate functions.
Hence, we will have shown that in order to ®nd the source-j adversarial rate function corresponding to each choice n;a we need only consider the rate functions in the set Sj. Further, since the rate functions inSjare characterized by a single parameter, Toff, this essentially involves a single- parameter optimization problem. We now proceed to formally state and prove the properties listed above.
Theorem 4. Every member of Sjis a feasible rate function.
Proof. Fix aToffand letbj tbe the corresponding member ofSj. It follows immediately from the de®nition ofbj tthat Zt
0bj sds#ej for all 0#t#Tj: 22
We can, in fact, show that Zt
0bj sds#ej t for allt$0: 23
To see this consider any arbitrarytnTj1s;wherenis some non-negative integer and 0#s#Tj:
Zt
0bj sdsZTj
0 bj sds1¼1ZnTj
n21Tj
bj sds
1ZnTj1s nTj
bj sds#nTjrLjj1ej s
# ej nTj1s2ej s1ej s ej t:
The ®rst inequality follows from Eq. (22) and from the fact that the average rate ofbj tover any period isrLjj:The second inequality follows because the slope ofej tis never less thanrLjj:
Also, because bj t is non-increasing over each of its periods, we have
Zt1t
t bj sds#Zt
0bj sds for allt$0; t$0: 24
Combining Eqs. (23) and (24) gives the desired result. B
Theorem 5. Each member of Sj maximizes MnUj a over the set of all feasible rate functions.
Proof. Each rate function inSjleads to the following form foruj t;the utilization of thejth segregated system:uj tis periodic with periodTj; and
uj t cnj 0#t#Don 0 Don#t#Tj; (
whereDonej Toff=cnj rLjj=cnjTj:
The corresponding steady-state random variable is
Uj
cnj with probability rLjj cnj ; 0 with probability 12 rLjj
cnj 0
@
1 A: 8>
>>
>>
<
>>
>>
>:
25
Note thatEUj rLjj:
For any feasible source, the steady state rate at which traf®c leaves the jth segregated system, U0j (say), must have a peak value less than or equal tocnj:Further, because the segregated system is lossless, the long-run average rate at which traf®c departs thejth segregated system must equal the long-run average rate at which traf®c enters the system, which is at most rLjj: Hence, we must have EU0j#rLjj: Among all random variables which have a peak value less than or equal tocnj and a mean value less than or equal torLjj; Ujas de®ned in Eq. (25) has the highest moment generating function,MUnj a:This is shown in the following argument (adapted from Ref. [18]). Let U0j be any non-negative random variable with distribution FU0j x with a peak valuec0#cnj and mean valuem0#rLjj:Then, sincea$0;
MUnj a2MUn0j a rLjj cnj
0
@ 1
Aeacnj 2 rLjj
cnj 112Zc0
0 eaxdFU0j x
$ m0 cnj
!
eacnj 2 m0 cnj 2Zc0
0 eax21dFU0j x
1 cnj
Zc0
0 x eacnj 212cnj eax21dFU0j x$0:
B Let bpj t be a rate function in Sj that has the largest MVnj a:
Theorem 6. bpj t maximizes MnVj a among all feasible rate functions.
Proof. Consider any feasible source-j rate function bj t
with periodTj. The actual arrival rate at time t is aj t bj t1uj where uj is the random phase. Here, we are concerned only with the steady-state distribution of the buffer contents and the utilization rate of thejth segregated system which are independent of the phase. Hence, in the rest of the proof, we will, without loss of generality, set the phase to zero and considerbj tto be the arrival rate at time t. The corresponding buffer contents process, vj t; is also periodic with periodTj.
In general, both bj t and vj t can have rather compli- cated forms with several intervals within a period where each is non-zero. However, we will ®rst show the desired result for feasible rate functions that give a buffer content process of the formvj t.0 for 0,t,tj andnj t 0 fortj#t#Tj;for some 0,tj,Tj:For rate processes of this form we have
vj t Zt
0bj sds2cnjt 0#t#tj; 0 tj#t#Tj: 8>
<
>:
Note that, sincevj t.0 for all 0,t,tj;we must have
tj#dj: 26
We show next thatMVnj acorresponding to such a feasi- ble rate function is smaller than that corresponding tobpj t:
We do this by showing that there is a rate function in the set Sj, bj t; with steady-state buffer contents Vj which is stochastically larger thanVjand which, hence, has a larger MGF (moment generating function).
LetToffbe such that ej Toff cjtj:From Eqs. (21) and (26) we get,ej Toff,ej tjiftj,djandej Toff ej djif tjdj: Hence, since ej ´ is non-decreasing and dj is uniquely de®ned, Toff #tj#dj: By de®nition, the rate function in Sj corresponding to this Toff is periodic with periodTjej Toff=rLjj and has the form
bj t e1j t 0#t#Toff; 0 Toff#t#Tj: 8<
:
The corresponding buffer contents at timet,yj t;is given as
yj t
ej t2cvjt 0#t#Toff; ej Toff2cvjt Toff #t#tj; 0 tj#t#Tj: 8>
><
>>
:
Denote the corresponding steady-state random variable as Vj:
Clearly,vj t#vj tfor all 0#t#Toff:Note, also, that we cannot havevj t.vj tfor anyToff #t#tjsince that would requirevj tto decrease at a rate strictly faster than cnj;in order for bothvj tandvj tto be zero attj. Hence, we get
nj t#nj t for all 0#t#tj: 27
Also, we can show that
Tj$Tj: 28
To see this, note that the utilization rate of thejth segre- gated system with arrival rate bj tis cnj whenever vj tis non-zero. Hence,P Ujcnj$tj=Tj:Also, since the aver- age utilization rate must be equal to the average arrival rate, which in turn is smaller thanrLjj;
rLjj$EUj$cvjP Ujcvj$cvj tj
Tj and so,
Tj$ cvjtj
rLjj ej Toff
rLjj Tj:
Eqs. (28) and (27) imply that
P Vj.x#P Vj.x for allx$0:
We have thus shown thatVjis stochastically smaller than Vjand hence has a smaller MGF. It is immediate from the de®nition of bpj tthatMnVj ais smaller than the MGF of bpj t:
We now extend this argument to the case of a general feasible rate functionbj t:Assume, without loss of gener- ality, that the corresponding buffer content process vj t
has m(some positive integer) non-zero portions within a single period, identi®ed by v1j;v2j;¼;vmj in the following manner:
vj t
0 0#t#tj1; v1j t2tj1 tj1#t#t1j 1t1j; v2j t2tj2 tj2#t#t2j 1t2j;
...
vmj t2tjm tmj #t#tmj 1tmj ; 0 tjm1tmj #t#Tj; 8>
>>
>>
>>
>>
>>
><
>>
>>
>>
>>
>>
>>
:
where tij.0; i1;2;¼;m; and tij$ti21j 1ti21j ; i 2;¼;m: Here, tij and tij1tij represent the end points of theith non-zero portion.
We can express each non-zero portionvij tas a periodic function, with periodTj, of the following form:
vij t Ztij1t
tij bj sds2cvjt 0#t#tij; 0 tij#t#Tj: 8>
><
>>
:
Let Vji denote the corresponding steady-state random