An evolutionary approach to inter-session network coding

Size: px
Start display at page:

Download "An evolutionary approach to inter-session network coding"

Transcription

1 An evolutionry pproch to inter-session network coding The MIT Fculty hs mde this rticle openly ville. Plese shre how this ccess enefits you. Your story mtters. Cittion As Pulished Pulisher Kim, M. et l. An Evolutionry Approch To Inter-Session Network Coding. INFOCOM 2009, IEEE IEEE Version Finl pulished version Accessed Fri Jn 26 23:47:07 EST 2018 Citle Link Terms of Use Detiled Terms Article is mde ville in ccordnce with the pulisher's policy nd my e suject to US copyright lw. Plese refer to the pulisher's site for terms of use.

2 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. An Evolutionry Approch To Inter-Session Network Coding Minkyu Kim, Muriel Médrd, Un-My O Reilly, Dnil Trskov Astrct Wheres the theory nd ppliction of optiml network coding re well studied for the single-session multicst scenrio, there is no known optiml network coding strtegy for more generl connection prolem where there re more thn one session nd receivers my demnd different sets of informtion. Though there hve een numer of recent studies tht demonstrte vrious utilities of network coding in the multisession scenrio, they rely on very restricted clsses of codes in terms of the coding opertions llowed nd/or the loction of decoding. In this pper, we propose novel inter-session network coding strtegy for generl connection prolem. Our coding strtegy llows firly generl rndom liner coding over lrge finite field, in which decoding is done t receivers nd the mixture of informtion t interior nodes is controlled y evolutionry mechnisms. We demonstrte how our coding strtegy my surpss existing end-to-end pirwise XOR coding schemes in terms of effectiveness nd prcticlity. I. INTRODUCTION In the multicst scenrio with single session, the theory of network coding is well founded on the fmous theorem y Ahlswede et l. [1] tht chrcterizes the cpcity region y the mx-flow (min-cut) ounds. Susequently, it is shown tht the optiml cpcity cn e chieved using only sclr liner network codes [2]. However, when we generlize the prolem such tht there re more thn one session nd receivers my demnd different sets of informtion, finding the optiml network coding strtegy is still n open question. First of ll, in such generlized prolem, chrcterizing the cpcity region ecomes prohiitively difficult nd even its inner/outer ounds cnnot e computed in prctice [3] [6]. Moreover, liner coding is shown to e insufficient for optiml coding in the multi-session cse [7]. A grph theoretic pproch is proposed s systemtic method for deciding solvility of given network with either liner or nonliner codes [8], whose sclility issue, however, is unresolved. Even within liner codes, the solvility decision prolem is shown to involve Gröner sis computtion [9], whose complexity my prohiit prcticl implementtions for lrge prolems. Recently, [10] suggests tht solving generl network coding prolem is equivlent in terms of complexity to solving set of polynomil equtions. M. Kim nd M. Médrd re with the Lortory of Informtion nd Decision Systems, Msschusetts Institute of Technology, Cmridge, MA 02139, USA ({minkyu, medrd}@mit.edu). U.-M. O Reilly is with the Computer Science nd Artificil Intelligence Lortory, Msschusetts Institute of Technology, Cmridge, MA 02139, USA (unmy@csil.mit.edu). D. Trskov is with the Institute for Communictions Engineering, Technicl University Munich, Munich, Germny (dnil.trskov@tum.de). In short, it is very unlikely tht network coding strtegy tht is theoreticlly optiml will soon emerge. Nevertheless, there hve een numer of studies tht demonstrte vrious utilities of network coding in the multi-session scenrio, sed on some restricted clsses of codes in terms of the coding opertions llowed nd/or the loction of decoding. Ktti et l. s opportunistic coding [11] leds to restricted, yet very prcticl coding scheme, tking dvntge of the rodcst nture of wireless medium. In their coding scheme, multiple, possily more thn two, pckets cn e comined using inry XOR, ut decoding needs to e done immeditely t the neighors of the coding node. If we wish to perform decoding t receivers, we my need to put n even stronger restriction: XOR opertions re llowed only etween two flows, thus clled pirwise XOR coding. The simplicity of coding nd decoding opertions of pirwise XOR codes llows the code construction prolem to e descried s flow formultions, which thus cn e solved jointly with vrious other network flow prolems. Trskov et l. [12] present liner nd integer optimiztion formultions for the clss of pirwise XOR coding for multiple unicst sessions. Ho et l. [13] develop ck pressure lgorithms for finding pproximtely throughput-optiml network codes within the clss of pirwise XOR coding. Eryilmz et l. [14] propose dynmic scheduling strtegy tht supports the chievle rtes for multiple unicst sessions given in [12]. Note, however, tht the enefit of pirwise XOR coding within tht frmework presented in [12], in terms of svings in link cost compred with trditionl routing, is found to e only modest depending on network topologies. It is not cler whether such modest gin is ecuse the simultions performed in [12] were restrictive, or it ctully indictes the limittions of the pirwise XOR coding. Even if we my wish to lift the restriction on the numer of sessions tht cn e coded together or the type of coding opertion, it is hrd to do so ecuse of the lck of pproprite tools for tht. The flow formultions in [12] or [13] my e generlized to represent more generl XOR coding scheme tht llows coding mong more thn two sessions. However, keeping trck of ll possile comintions of coded strems mong up to k sessions would require t lest O(m k E V ) vriles, where m is the totl numer of sessions, which leds to prohiitively lrge numer of constrints, hindering prcticl implementtions. Beyond the inry field, pthsed chrcteriztion is presented for the fesiility of the connection prolem with two unicst sessions [15], sed on which distriuted rte control lgorithm is proposed /09/$ IEEE 450 Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply.

3 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. [16]. However, generlized chrcteriztion in the cse of coding mong more thn two sessions still seems difficult. For more generl coding schemes thn the pirwise XOR tht llow decoding t receivers, liner optimiztion prolem is proposed y Lun et l. [17], whose minimum cost is shown to e no greter thn the minimum cost of ny routing solution. However, its formultion is sed on the given set of the source processes tht cn e mixed on ech link, which still remins difficult to decide optimlly. In this pper, given tht there exists wide gp etween the presumly difficult quest for optiml inter-session network coding nd mny existing pproches tht re constrined within very restrictive clsses of coding strtegies, we investigte how the evolutionry pproches sed on Genetic Algorithm (GA), which in our previous works [18] [22] hve een used for the network coding resource optimiztion prolem, cn e utilized in n effort to fill the gp. In prticulr, we present novel rndomized liner coding scheme, in which decoding hppens t receivers while interior nodes perform rndom liner coding with selective mixture of informtion, controlled y evolutionry mechnisms. We then demonstrte, through simultions, how our coding strtegy my surpss existing end-to-end XOR coding schemes in terms of effectiveness nd prcticlity. The rest of the pper is orgnized s follows. Section II descries the prolem setup with relted ckground. Section III presents our coding strtegy with rief introduction to GA. Section IV descries how the computtionl components of our evolutionry pproch need to e designed, nd then Section V shows how our pproch cn e implemented in distriuted fshion over the network. Section VI exhiits numer of simultions for the evlution of the performnce of our coding strtegy. Section VII concludes the pper with some directions for future reserch. A. Prolem Setup II. PROBLEM SETUP AND BACKGROUND We ssume tht the network is given y n cyclic directed multigrph G =(V,E) where ech link hs unit cpcity nd links with lrger cpcities re represented y multiple links. Only integer flows re llowed, hence there is either no flow or unit rte of flow on ech link. Let us ssume tht there re R independent rndom processes of unit entropy rte, denoted y X ={X 1, X 2,..., X R }, originting t s( 1) source nodes. There re d receiver nodes, v 1, v 2,..., v d. For ech receiver v i (i =1,..., d), we denote the set of requested source processes y X i X. We ssume tht for ech source process in X i (i =1,..., d), there exists pth from its originting node to receiver v i ; otherwise, it is esy to check nd declre tht the prolem is not solvle. Other thn this connectivity condition, we do not put ny restriction on the source processes ech receiver node my request, i.e., X i cn e ny nonempty suset of X provided tht receiver v i is rechle from the originting nodes of the source processes in it. For this generlized scenrio, there is no simple chrcteriztion known for the fesiility of the connection prolem with network coding, even for the simpler cse of unicst connections, i.e., when ll X i s re disjoint [9], [23]. Though liner coding hs een shown to e suoptiml in generl [7], here we focus on the sclr liner coding in generl form, i.e., we do not restrict ourselves within either inry XOR opertions or pirwise mixing. Becuse the fesiility chrcteriztion still remins hrd even within liner coding [9], we ssume tht the cpcity constrints cn e relxed for some links, i.e., links cn e scled up, if necessry, to llow multiple trnsmissions. Note tht this my e the cse in mny prcticl networking scenrios; e.g., in wireless networks, nodes cn exploit more cpcities t the expense of more energy, nd in opticl networks, cpcity itself is rrely limited resource, though using more resources incurs n dditionl cost. With this ssumption, the fesiility prolem cn e resolved y scling links ppropritely. Then, we minly focus on finding cost-efficient trnsmission scheme using network coding. Hence, our primry ojective is to minimize the link cost required to stisfy the given communiction demnds. Lter, we consider nother ojective of minimizing the coding resources, i.e., numer of coding nodes/links. B. Bckground For the multicst cse, the solvility of connection prolem oils down to whether the mx-flows etween the source node nd ech of the receiver nodes ll exceed the desired multicst rte [1]. This cn e trnslted into the lgeric frmework [9] such tht, if we let ξ denote the vector consisting of ll the link coefficients s defined in [1], the prolem of finding fesile network code ecomes finding n ssignment of numers to vriles ξ such tht the product of the determinnt polynomils does not evlute to zero, which cn e done reltively esily with mny efficient rndomized lgorithms, e.g., s in [24]. As we go eyond the multicst cse, we now hve nother condition for solvility: within the trnsfer mtrix, the sumtrices relting the input processes tht re not requested to the corresponding output processes t the receiver nodes should evlute to zero (the first condition in Theorem 6 in [9]). Let us denote the entries of the sumtrices tht hve to evlute to zero y f 1 (ξ),f 2 (ξ),..., f K (ξ), which we refer to s interference polynomils. As in the multicst cse, we still hve the condition tht the sumtrices ssocited with the input nd output processes specified in the connection requests should e nonsingulr (the second condition in Theorem 6 in [9]). Let g 1 (ξ),g 2 (ξ),..., g L (ξ), referred to s determinnt polynomils, e the determinnts of the sumtrices tht should e nonsingulr. While finding n ssignment tht mkes determinnt polynomils nonzero cn e done esily, the difficult prt is to solve the set of determinnt polynomils to find n ssignment of numers to ξ tht mkes them ll zero. To determine whether Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 451

4 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. such n ssignment is possile, [9] suggests Gröner sis method for determining the emptiness of vriety of the idel generted y the interference polynomils nd function of the determinnt polynomils. However, the worst cse complexity of the Bucherger lgorithm, which is used predominntly to compute Gröner sis, is douly exponentil [25] nd moreover, in our cse the input to the Bucherger lgorithm, i.e., the interference nd determinnt polynomils, my lredy hve n exponentil numer of terms, nd hence clculting Gröner sis my not e prcticl solution nywy. On the other hnd, s solving generl liner network coding prolem is no esier thn solving set of polynomil equtions [10], it is unlikely tht one cn develop specilized method tht possily provides more efficiency, utilizing some structurl properties of the interference nd determinnt polynomils. III. OUR CODING STRATEGY: SELECTIVE RANDOM LINEAR CODING In our coding strtegy, we control the mixture of informtion t ech node y deciding whether to include the input from prticulr incoming link when clculting the output on ech outgoing link. However, once the decisions re mde, we clculte the output y forming rndom liner comintion of the inputs llowed. Hence, the nme selective rndom liner coding. Intuitively, the strtegy we employ to del with the interference polynomils is 1) to mke some interference polynomils identiclly zero y zeroing out n enough numer of ssocited components of ξ nd 2) for the interference polynomils tht remin nonzero, to llow enough degrees of freedom t the receivers so tht the unwnted informtion cn e successfully cnceled out. For node v with d in incoming links nd d out outgoing links, we ssign inry vrile ij to ech pir of the i {1,..., d in }-th incoming link nd the j {1,..., d out }-th outgoing link. For the j-th (j =1,..., d out ) outgoing link, we refer to the ssocited inry vriles j =( ij ) i {1,...,din} s coding vector (see Fig. 1 for n exmple). The coding vectors re the vriles tht we need to decide in our coding strtegy. x1 x2 x3 y1 v y2 () Node v x1 x2 x3 v x1 x2 x3 v y1 y2 1= = coding vector for y1 coding vector for y2 () Coding vectors for outgoing links Fig. 1. Node v with 3 incoming nd 2 outgoing links is ssocited with two coding vectors 1 =( 11, 21, 31 ) nd 2 =( 12, 22, 32 ). Once set of the coding vectors is given, we employ rndom liner coding t interior nodes, selectively using only those inputs ssocited with 1 s in the coding vectors. More specificlly, node v clcultes the output y j (j =1,..., d out ) on the j-th outgoing link s follows. Define the set I j of indices s I j = {1 i d in the i-th component of j is 1}. If we denote the input from the i-th (i =1,..., d in ) incoming link y x i, y j = i I j rnd(f q ) x i where rnd(f q ) denotes nonzero rndom element from F q. If the set I j is empty, y j is ssumed to e zero. Given set of coding vectors, the fesiility verifiction cn e done y first performing selective rndom liner coding t interior nodes s descried ove. Then, ech receiver node v i (i =1,..., d) performs Gussin elimintion to determine whether ll the desired input processes cn e recovered, with the interference prt cnceled out. Note tht this whole process cn e done in distriuted mnner, which will e utilized lter. This rndomized decision rule incurs n error when nonzero polynomils evlute to zero fter rndomly ssigning vlues to vriles; for zero polynomils, rndom ssignments in the evlution process do not ffect the finl result. Hence, the error proility is ounded y 1 (1 d/q) ν where q is the size of the finite field used for coding nd ν is the mximum numer of links in ny set of links constituting flow solution from the source to ny receiver [24]. Note tht this ound remins the sme even if we scle up some of the links s will e discussed lter. It remins to find n optiml ssignment of the coding vectors out of exponentilly scling numer of possile choices, which we ddress using GA sed serch method. Before proceeding, let us provide rief introduction to GA. A. A Brief Introduction to GA GAs re stochstic serch methods tht mimic genetic phenomen such s gene recomintion, muttion nd survivl of the fittest. Hving een pplied to lrge numer of scientific nd engineering prolems, GAs re especilly shown to e effective for the prolem of network coding resource optimiztion [18] [22]. The min control flow of the stndrd form of GA, clled simple GA, is shown in Fig. 2 [26]. Simple GA [26] opertes on set of cndidte solutions, clled popultion. Ech solution is typiclly represented y it string, clled chromosome. Ech chromosome is ssigned fitness vlue tht mesures how well the chromosome solves the prolem t hnd, compred with other chromosomes in the popultion. From the current popultion, new popultion is generted typiclly using three genetic opertors: selection, crossover nd muttion. Chromosomes for the new popultion re selected rndomly (with replcement) in such wy tht chromosomes tht re more fit re selected with higher proility. For crossover, chromosomes re rndomly pired, nd then two chromosomes in ech pir exchnge suset Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 452

5 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. initilize popultion; evlute popultion; while termintion criterion not reched { select solutions for next popultion; perform crossover; perform muttion; evlute popultion; } Fig. 2. Min control flow of simple GA. of their it strings to crete two offspring. Chromosomes re then suject to muttion, which refers to rndom flips of the its pplied individully to ech of the new chromosomes. The process of evlution, selection, crossover nd muttion forms one genertion in the execution of simple GA. The ove process is iterted with the newly generted popultion successively replcing the current one. Simple GA termintes when certin stopping criterion is reched, e.g., fter predefined numer of genertions. For further detils of stndrd simple GA, the reder is referred to [18], [26]. IV. COMPUTATIONAL ASPECTS We first descrie how the computtionl components of our evolutionry pproch need to e designed, nd then in the next section, we present how our pproch cn e implemented in distriuted mnner over the network. A. Chromosome nd Fitness Function The decision vriles in our coding scheme re set of coding vectors s defined ove in Section III. Hence, ech chromosome (i.e., cndidte solution) consists of the collection of ll coding vectors. For given chromosome y,we must verify its fesiility first y performing selective rndom liner coding s descried in Section III. Once the fesiility test is done, given tht our primry ojective is to otin the cost-efficient trnsmission scheme vi network coding, the fitness function F is defined s { totl cost of link usge, if y is fesile, F (y) = (1), if y is infesile. Note tht if we wish to minimize the resources engged in network coding, we my replce the link cost y the numer of coding nodes/links for fesile chromosomes. Moreover, the two ojective vlues, i.e., the link nd coding costs, cn e jointly considered to investigte possile trdeoff etween those two ojectives [22], which leds to more informed decisions on whether or where to employ network coding. B. Initil Popultion Construction Typiclly the initil popultion of GA is composed of rndom chromosomes. However, s pointed out in [18], inserting some non-rndom chromosomes cn gretly improve the performnce of the lgorithm. When we dd numer of non-rndom solutions to the initil popultion, cre must e tken to insert only neutrl solutions in the sense tht they re not prticulrly close to some locl optimum. Otherwise, those inserted strting points tend to tke over the whole popultion in the erly stge of the evolution process so tht the lgorithm my end up converging just to neighorhood of the strting points. To crete neutrl strting point, we first scle up the links in G to mke the prolem multicst-like in the sense tht ech receiver node now receives ll source processes tht hve directed pth to it. Then, we employ s neutrl strting point the ll-one chromosome tht indictes mixing everything t ll interior nodes. Let us now show how we cn find n pproprite scling fctor. Before proceeding, for ech receiver node v i (i = 1,..., d) in G, welety i X e the set of the source processes from whose originting node, receiver v i is rechle (i.e., there exists directed pth from the source process s originting node to node v i ). Also, we let r i (i =1,..., d) denote the size of set Y i. Now let us crete n uxiliry network H from G y introducing virtul source node S nd dding unitcpcity link from node S to the source node t which ech source process X j X(j =1,..., R) origintes. Then, we let f i (i =1,..., d) e the vlue of the mximum flow from the virtul source node S to ech receiver v i (i =1,..., d). We define k s k = mx i {1,...,d} ( ri f i ). (2) Theorem 1: Let G e scled version of G with ech link replced y k multiple links, where k is defined s in (2). In network G, network code tht performs rndom liner coding using ll ville inputs t every interior node in sufficiently lrge finite field is fesile. Proof Outline: Let us crete nother uxiliry network H y scling up the links in H y fctor of k nd dding unit-cpcity links etween ech source process in X\Y i nd ech receiver node v i (i = 1,..., d). Then, we hve multicst prolem for which fesile network code cn e otined y employing rndom liner coding in sufficiently lrge finite field. Given such fesile network code C on network H, it cn e shown tht C remins fesile even if we remove the dded links to otin network G.(Afull proof cn e found in [27].) Note tht the ll-one chromosome shown ove is fesile ut hs the worst cost in terms of either the link cost or the coding cost nd thus it my not is the initil popultion towrd ny prticulr suoptiml solution. C. Genetic Opertors In conventionl crossover nd muttion opertors, which we refer to s it-wise opertors, the unit of interchnged or pertured sucomponents is ech it of the chromosomes. However, it is pointed out in [19], [20] tht, for the prolem of the coding resource optimiztion for multicst, significnt Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 453

6 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. performnce gin cn e ttined y pplying the crossover nd muttion opertions on ech full coding vector. Tht is, for vector-wise crossover, we let two chromosomes suject to crossover exchnge ech full coding vector (rther thn ech it) independently with the given crossover proility. For vector-wise muttion, we rndomly regenerte ech coding vector (gin, rther thn ech it) independently with the given muttion proility. We refer to these two opertors s vector-wise opertors. Given tht enforcing coding vector-level modulrity in genetic opertions cn led to significnt performnce gin, we my develop nother set of genetic opertors tht further exploit, this time, the node-level modulrity. In node-wise opertors, we now interchnge or pertur ll coding vectors ssocited with the outgoing links of ech node with the crossover/muttion proility. The intuition ehind the nodewise genetic opertors is tht the coefficients of the links directly connected with ech other or within few hops wy re more likely to hve strong dependencies thn those ssocited with the links fr wy from ech other. Note, however, tht the node-wise genetic opertors enforce roder level of modulrity thn the vector-wise opertors, exchnging or perturing lrger numer of coding vectors t once. If this is the right level of modulrity, it would trnslte into fster convergence of the lgorithm to the sme or etter solution; otherwise, the lgorithm my tend to converge premturely to lower qulity solution. We will lter verify through simultions the effect of these different genetic opertors on the performnce of the lgorithm. V. DISTRIBUTED IMPLEMENTATION Recll tht in Section III we discussed how the fesiility test of single chromosome cn e done y employing rndom liner coding t interior nodes. Note tht in doing so, ech interior node only refers the relevnt portion of the chromosome, i.e., the coding vectors tht indicte the opertions t tht node. Hence, we cn divide up the popultion y letting ech node hndle only the coding vectors it needs from every chromosome in the popultion. If the coding vectors re stored t locl nodes, we only need to trnsmit the set of the coefficients tht indicte the overll effect of network coding reltive to the source dt, which is commonly referred to s glol encoding vector in the network coding literture (see e.g., [28]), for fesiility test. Hence, fesiility test cn e done in distriuted fshion y trnsmitting pckets contining such coefficients. Moreover, it cn e shown tht the whole fitness clcultion nd ll genetic opertions discussed in Section IV cn e done independently t locl interior nodes with some coordintion informtion emedded in dt pckets, ssuming tht the source nodes cn communicte with one nother [27]. Note tht when we consider network coding mong multiple sessions tht my originte from multiple source nodes, it must e ssumed in the first plce nywy tht the source nodes cn communicte with one nother to find out the totl numer of the source processes eing considered together for possile coding so tht the coefficients for liner coding cn e ligned consistently cross ll the source nodes involved. Hence, we ssume tht mong the prticipting source nodes, we cn designte one s the mster node which serves s the min controller of the lgorithm y gthering informtion from other source nodes nd sending the clculted coordintion informtion to other source nodes. With the ove ssumptions, the whole evolutionry lgorithm to serch for n optiml set of coding vectors cn operte in distriuted fshion [27], whose overll flow is shown in Fig. 3 with the loction of ech procedure specified. More detiled descriptions of the distriuted lgorithm re omitted for spce considertion (the reder is referred to [27, Chpter 7] for detils). Note tht the clcultion of the scling fctor k defined in (2) cn lso e done in distriuted mnner [27]. Computtionlly, the distriuted lgorithm performs the sme tsk s simple GA (Fig. 2) with the computtionl components descried in Section IV. [S1] initilize; (ll nodes) [S2] run forwrd evlution phse; (ll nodes) [S3] run ckwrd evlution phse; (ll nodes) [S4] send prtil fitness to mster node; (source nodes) [S5] clculte fitness; (mster node) [S6] while termintion criterion not reched (mster node) { [S7] clculte coordintion vector; (mster node) [S8] fetch coordintion vector from mster node; (source nodes) [S9] run forwrd evlution phse; (ll nodes) [S10] perform selection, crossover, muttion; (interior nodes) [S11] run ckwrd evlution phse; (ll nodes) [S12] [S13] Fig. 3. send prtil fitness to mster node; (source nodes) clculte fitness; (mster node) } Flow of distriuted lgorithm for selective rndom liner coding Within this distriuted setup, the fitness evlution of ech chromosome requires the computtionl complexity of O(d in d out R) t ech interior node nd O(d 2 in R) t ech receiver node [27]. The most importnt enefit of this distriuted structure is tht network code cn e constructed on the fly while the network is opertionl, llowing for the following network coding protocol: As the mster node sends pcket tht signifies the strt of the code construction, ll prticipting nodes go into the code construction mode, running the lgorithm descried ove. As the distriuted evolutionry lgorithm proceeds, ech interior node stores nd improves its relevnt network codes. At the end of the lgorithm, the mster node only needs to send the index of the est chromosome of the lst popultion nd ll prticipting nodes now strt to trnsmit dt sed on the loclly stored network code tht corresponds to the received est index. Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 454

7 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. VI. PERFORMANCE EVALUATION In order to evlute the effect of different genetic opertors nd lso compre the performnce of our coding scheme with others, we first perform simultions for the multipleunicst scenrio, for which other network coding schemes re ville. Then, to demonstrte our lgorithm s ility to hndle more generl prolems, we further consider the cse of wireless networks with no restriction on the type of connection requests. A. Multiple Unicst Connections We performed numer of simultions for the multipleunicst cse, i.e., ll connection requests X i s (i =1,..., d) re disjoint. Our simultions re sed on the grid network introduced in [12] (network D in Fig. 4). In [12], the cost of ech link is ssigned rndomly, which llows only slim chnces tht the network coding dvntge exists. Note, however, tht network coding gin tkes effect only if there exists n expensive ottleneck link tht hs to e used y numer of flows nd lso lower cost detours round it, which hppens rrely when the connection requests nd link costs re rndomly chosen. Hence, in our experiments, we pick cost ssignment s depicted in Figure 4, where the links with cost higher thn 1 re highlighted y thicker rrows with the ctul cost shown y their side, to mke the network coding dvntge clerly exist t lest for some connection requests. For comprison, we tke Trskov et l. s pirwise XOR coding scheme [12]. We do not include the pproch y Wng et l. [15] here ecuse it cn e shown tht the gril structure considered therein does not led to link cost svings in our prolem setup (i.e., our primry ojective to minimize the link cost wheres the cpcity constrint my e relxed if needed); nevertheless, it does increse the cpcity region when the cpcity constrint is strictly enforced. Fig u1 u2 u3 v x y v2 Grid network D for multiple-unicst simultions 1) Two-Connection Cse: First, we consider the cse of two connections, i.e., R =2. We repetedly rn our lgorithm, vrying the loction of the source processes nd connection requests such tht two source processes X 1 nd X 2 my originte t ny of nodes {u 1,u 2,u 3 } nd ech source process v my e requested t ny of receiver nodes {v 1,v 2,v 3 }.For comprison with the routing cse nd the pirwise XOR coding scheme, we used the multi-commodity formultion with integer constrints nd Trskov et l. s formultion [12], respectively. Among ll possile rrngements of the source processes nd connection requests, there is only one cse in which network coding sves the link cost: X 1 nd X 2 originte t nodes u 1 nd u 3, respectively, while X 1 nd X 2 re requested t nodes v 3 nd v 1, respectively. Fig. 5() shows the est solution otined y our selective rndom liner coding which requires the link cost of 131. Note tht, for this simple prolem, our evolutionry lgorithm, despite its stochstic nture, lwys yields the sme solution regrdless of the genetic opertors used. In comprison, the optiml cost chieved y routing is 212. In Fig. 5(), we use the symol to represent liner comintion with rndom coefficients from the designted finite field; i.e., x 1 x 2... x n = n i=1 rnd(f q) x i, where gin rnd(f q ) represents nonzero rndom element from the designted finite field F q. Interestingly, this exmple highlights the difference etween our coding scheme nd the pirwise XOR coding scheme, whose est solution, s depicted in Fig. 5(), offers n even lower cost of 129. In our selective rndom liner coding, coding opertion is performed only once t node x nd decoding is done t receiver nodes. On the other hnd, in the est pirwise XOR code, nother XOR opertion is done t node z, which, in fct, serves s decoding t n interior node nd consequently sves the link cost of 2 (y not sending β t node w long longer pth down to node u). Note tht the sme trnsmission strtegy would not work for our selective rndom liner coding ecuse nother coding opertion t node z would yield nother rndom liner comintion ( ) rther thn. From this exmple, we oserve tht, while oth coding schemes provide dvntges over trditionl routing, our selective rndom liner coding scheme my require some dditionl link cost compred with the pirwise XOR coding, which cn e considered the price of employing rndomized coding in finite field lrger thn the inry field. Note, however, tht s we increse the numer of connections, our coding scheme leds to much more prcticl solution despite the possile expense of such dditionl link cost, s will e discussed next. 2) Five-Connection Cse: Let us now increse the numer of source processes to 5 nd pick up n rrngement of the source processes nd connection requests tht llows the network coding dvntge: {(source process, source node, receiver node)} = {(X 1,u 1,v 3 ), (X 2,u 1,v 1 ), (X 3,u 3,v 1 ), (X 4,u 3,v 2 ), (X 5,u 3,v 2 )}. As opposed to the simpler prolem in the previous susection, the performnce of our lgorithm vries depending on the type of genetic opertors used. Tle I summrizes the performnce of our lgorithm with different genetic opertors. For ech type of genetic opertors, we performed 30 simultion runs nd t the end of ech run we pick the est coding solution out of the lst popultion. The first column shows the lowest cost of those 30 est coding solutions nd the next Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 455

8 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. u1 u2 u3 u1 u2 u3 x x y w z y z w v1 v2 v3 v1 v2 v3 () Selective rndom liner coding () Pirwise XOR coding Fig. 5. Comprison of est coding solutions otined. two columns show the verge nd stndrd devition of the cost of those 30 solutions. For comprison, the est routing solution yields the optiml cost of 242. The fourth column of the tle displys the rtio of the lgorithm runs, out of the totl of 30, in which the cost of the est coding solution found is ctully lower thn the est routing solution. The fifth column clcultes the verge numer of genertions required for coding solution tht outperforms the routing solution, if found, to pper in the popultion. The lst column shows the rtio of the lgorithm runs in which coding solution outperforming the routing solution is found efore the 100- th genertion, reltive to the totl numer of lgorithm runs where such coding solution is ever found. TABLE I SUMMARY OF THE LINK COSTS OF THE CODING SOLUTIONS FOUND BY DIFFERENT GENETIC OPERATORS AND RELATED STATISTICS. THE BEST ROUTING SOLUTION REQUIRES THE LINK COST OF 242. Link Cost Outperform Routing Best Avg Std Rtio At <100 Bit-wise Vector-wise Node-wise From Tle I, we first notice tht our lgorithm with the itwise or vector-wise opertors relily yields network coding solution tht outperforms the est routing solution. Between the two kinds of opertors, the vector-wise opertors led to much etter solutions, oth in terms of the men nd stndrd devition. Also, the numer of genertions required to find network coding etter thn the est routing solution is much smller on verge for the vector-wise opertors; most of the time it is found efore the 100-th genertion. Hence, we my conclude tht the vector-wise opertors llow the lgorithm to find much etter solutions much fster thn the it-wise opertors. For the cse of the node-wise opertors, our lgorithm finds network coding solution tht exceeds the routing solution only in 43% of the simultions. However, in those successful simultions, such network coding solution is found much fster thn the cse of the vector-wise opertors. Hence, we my conclude tht with the node-wise opertors the lgorithm tends to converge fster, ut premturely to lower qulity solution. This my e due to tht the node-level modulrity enforced y the node-wise opertors is too strong in the sense tht it chnges too mny coding vectors t once, which conceptully my correspond to setting the step too lrge in n itertive optimiztion scenrio. On the other hnd, the optimiztion formultions for the pirwise XOR coding [12] filed to converge within resonle mount of time (during full week of simultions) sed on the simultion environment used in [12]. Note tht the liner nd integer progrms in [12], even for this fiveconnection prolem, contin round nd 1400 vriles (including the slck vriles to hndle the mx opertor in the constrints) nd nd 1700 constrints, respectively. Though we my hve een le to otin converged results if we hd experimented it with much fster mchine, the point we would like to mke here is tht the optimiztion formultions considered in [12] my not provide prcticl sclility s the numer of connections increses. For comprison, our lgorithm tkes out 1.5 seconds for ech genertion in the sme simultion environment, where we simulted ech node s opertion sequentilly (one t ech time from upstrem to downstrem nodes) using MATLAB on single-processor mchine. In rel network, where ech node uses its own computtionl resources, we my expect much fster execution of our lgorithm. B. Generl Connections We performed nother set of experiments in fully generlized networking scenrios where the receiver nodes my request ny comintion of the ville source processes. For this, we generted 100 rndom wireless networks where 40 nodes re plced rndomly within squre with rdius of connectivity 3. A unit-rte hyperlink is originted from ech node towrd the set of nodes tht re within the Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 456

9 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. connectivity rnge nd hve higher horizontl coordinte. In ech rndom topology, 5 source processes were rndomly plced t 5 nodes chosen in the left hlf nd ech of 5 receiver nodes rndomly chosen in the right hlf demnds rndomly chosen nonempty suset of X = {X 1,..., X 5 }. For such generted rndom connection prolems, we pply two-stge method in which we first run our lgorithm with the fitness function defined s (1) to minimize the link cost without restricting network coding. Then, from the est coding solution found, we tke only the links tht re used for trnsmission to form the sugrph to e used in the second stge. In the second stge, we use our lgorithm to minimize the numer of coding nodes y chnging the fitness function to reflect the numer of nodes where network coding is performed. Tle II shows the distriution of the found minimum numer of coding nodes through the two-stge method. Note tht in most cses (out 80% of the rndom topologies tested), the clculted minimum cost cn e chieved without network coding t ll. Also, for the topologies where network coding is required, it is required only t smll suset of nodes, i.e., not ll nodes need to perform network coding, which ws lso the cse in mny multicst scenrios [27]. TABLE II DISTRIBUTION OF THE CALCULATED MINIMUM NUMBER OF CODING NODES IN 100 RANDOM TOPOLOGIES. # Coding Nodes # Topologies C. Discussion Aove simultions exhiit tht our evolutionry pproch offers coding strtegy in the generl connection prolem which is still somewht restricted ut with enhnced fetures in mny spects, compred with existing pirwise XOR coding. For the prolem of multiple unicst connections, we showed tht our evolutionry pproch pproch yields network coding solution tht offers n dvntge over trditionl routing in terms of link cost, wheres n existing pproch for pirwise XOR coding my fil to provide sclility within prcticl rnges s the numer of connections increses. Also, we demonstrted tht our evolutionry pproch cn tckle more generl prolems in which there is no restriction on the type of connection requests, tking the coding cost into ccount s well. Note tht, though only the two-stge method ws experimented for n illustrtive purpose in the second set of simultions, multi-ojective evolutionry pproch cn lso e utilized, similrly s in [22], to investigte the trdeoff etween the two ojectives. In ddition to the distriuted structure presented in Section V, nother distriuted structure, i.e., temporlly distriuted structure, cn lso e dopted for more efficient utiliztion of computtionl resources s well s more roust opertion ginst pcket losses [21]. A possile drwck of our evolutionry pproch is tht, s opposed to the cse to the coding resource optimiztion for multicst [19], it lcks performnce ound. However, given tht there re no prcticl lterntives tht tke into ccount coding mong more thn two flows, our pproch my serve s unique mens for exploring network coding dvntges in much more generlized setup thn the pirwise coding. Another possile limittion is regrding the sclility; i.e., s is typicl for GA, the popultion size my need to e incresed significntly for lrge prolems. However, with the distriuted structure descried in Section V, the popultion size cn e incresed rther flexily y incresing the size of the pcket used for fitness evlution, given tht the computtionl complexity t ech node scle linerly with the popultion size. Moreover, with the temporlly distriuted structure [21], we my further increse the effective popultion size y incresing the numer of (su)popultions. VII. CONCLUSION AND FUTURE WORK We hve proposed novel inter-session network coding strtegy for generl connection prolem eyond multicst, for which no optiml network coding scheme is known. Our coding strtegy llows rndom liner coding over lrge finite field, in which decoding is done t receivers nd the mixture of informtion t interior nodes is controlled y evolutionry mechnisms. We hve demonstrted how our coding strtegy my surpss existing end-to-end pirwise XOR coding schemes in terms of effectiveness nd prcticlity. There hve een mny recent developments in the field of evolutionry computtion tht significntly improve the sclility of the trditionl simple GA, on which our current pproch is sed. Such dvnced GA frmeworks cn e redily employed, without chnging the overll frmework presented here, to find nd exploit further linkge informtion, i.e., dependencies mong vriles, for n improved performnce. Also, one my develop method to construct solutions tht provide some useful ounds on the link or coding cost, which in turn cn e comined with our evolutionry pproch to generte etter initil popultion or to refine the finl solution (s in the coding resource optimiztion prolem for multicst [19]). In ddition, one my utilize the core structure of our proposed evolutionry pproch within other network prolems tht involve comintoril optimiztions with n unresolved scling issue. For instnce, for prolem in which the numer of vriles/constrints scles exponentilly with the size of the network, n evolutionry lgorithm my effectively e utilized to yield more compct set of vriles/constrints tht needs to e considered t time. REFERENCES [1] R. Ahlswede, N. Ci, S.-Y. R. Li, nd R. W. Yeung, Network informtion flow, IEEE Trns. Inform. Theory, vol. 46, no. 4, pp , [2] S.-Y. R. Li, R. W. Yeung, nd N. Ci, Liner network coding, IEEE Trns. Inform. Theory, vol. 49, no. 2, pp , [3] R. W. Yeung, S.-Y. R. Li, N. Ci, nd Z. Zhng, Network coding theory prt II: Multiple source, Foundtion nd Trends in Communictions nd Informtion Theory, vol. 2, no. 5, pp , [4] R. W. Yeung, A First Course in Informtion Theory,. Kluwer Acdemic/Plenum Pulishers, Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 457

10 This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2009 proceedings. [5] T. Chn nd A. Grnt, Dulities etween entropy functions nd network codes, sumitted to IEEE Trns. Inform. Theory (rxiv: v1), [6] X. Yn, R. Yeung, nd Z. Zhng, The cpcity region for multi-source multi-sink network coding, in Proc. IEEE ISIT, [7] R. Dougherty, C. Freiling, nd K. Zeger, Insufficiency of liner coding in network informtion flow, IEEE Trns. Inform. Theory, vol. 51, no. 8, pp , [8] J. K. Sundrrjn, M. Médrd, R. Koetter, nd E. Erez, A systemtic pproch to network coding prolems using conflict grphs, in Proc. Informtion Theory nd its Applictions, [9] R. Koetter nd M. Médrd, An lgeric pproch to network coding, IEEE/ACM Trns. Networking, vol. 11, no. 5, pp , [10] R. Dougherty, C. Freiling, nd K. Zeger, Liner network codes nd systems of polynomil equtions, IEEE Trns. Inform. Theory, vol. 54, no. 5, pp , [11] S. Ktti, D. Kti, W. Hu, H. Rhul, nd M. Médrd, The importnce of eing opportunistic: Prcticl network coding for wireless environments, in Proc. Allerton Conference, [12] D. Trskov, N. Rtnkr, D. S. Lun, R. Koetter, nd M. Médrd, Network coding for multiple unicsts: An pproch sed on liner optimiztion, in Proc. ISIT, 2006, pp [13] T. Ho, Y.-H. Chng, nd K. J. Hn, On constructive network coding for multiple unicsts, in Proc. Allerton Conference on Communiction, Control nd Computing, [14] A. Eryilmz nd D. S. Lun, Control for inter-session network coding, MIT-LIDS, Tech. Rep. P-2722, [15] C.-C. Wng nd N. B. Shroff, Beyond the utterfly - grph-theoretic chrcteriztion of the fesiility of network coding with two simple unicst sessions, in Proc. IEEE ISIT, [16] A. Khreishh, C.-C. Wng, nd N. B. Shroff, An optimiztion sed rte control for communiction networks with inter-session network coding, in Proc. IEEE Infocom, [17] D. S. Lun, M. Médrd, T. Ho, nd R. Koetter, Network coding with cost criterion, MIT-LIDS, Tech. Rep. P-2584, [18] M. Kim, C. W. Ahn, M. Médrd, nd M. Effros, On minimizing network coding resources: An evolutionry pproch, in Proc. NetCod, [19] M. Kim, M. Médrd, V. Aggrwl, U.-M. O Reilly, W. Kim, C. W. Ahn, nd M. Effros, Evolutionry pproches to minimizing network coding resources, in Proc. IEEE Infocom, [20] M. Kim, V. Aggrwl, U.-M. O Reilly, nd M. Médrd, Genetic representtions for evolutionry minimiztion of network coding resources, in Proc. EvoWorkshops, [21], A douly distriuted genetic lgorithm for network coding, in Proc. ACM Genetic nd Evolutionry Computtion Conference (GECCO), [22] M. Kim, M. Médrd, V. Aggrwl, nd U.-M. O Reilly, On the codinglink cost trdeoff in multicst network coding, in Proc. MILCOM, [23] Z. Li nd B. Li, Network coding: The cse of multiple unicst sessions, in Proc. Annul Allerton Conference on Communiction, Control, nd Computing, [24] T. Ho, M. Médrd, R. Koetter, D. R. Krger, M. Effros, J. Shi, nd B. Leong, A rndom liner network coding pproch to multicst, IEEE Trns. Inform. Theory, vol. 52, no. 10, pp , [25] M. Milnič, Report on DIMACS working group on dt de-identifiction, comintoril optimiztion, grph theory, nd the stt/or interfce, [26] M. Mitchell, An Introduction to Genetic Algorithms. MIT Press, [27] M. Kim, Evolutionry pproches towrd prcticl network coding, Ph.D. disserttion, MIT, 2008, (ville online t [28] P. A. Chou, Y. Wu, nd K. Jin, Prcticl network coding, in Proc. Annul Allerton Conference on Communiction, Control, nd Computing, Authorized licensed use limited to: MIT Lirries. Downloded on April 09,2010 t 16:34:56 UTC from IEEE Xplore. Restrictions pply. 458

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming Lecture 10 Evolutionry Computtion: Evolution strtegies nd genetic progrmming Evolution strtegies Genetic progrmming Summry Negnevitsky, Person Eduction, 2011 1 Evolution Strtegies Another pproch to simulting

More information

Evolutionary Approaches To Minimizing Network Coding Resources

Evolutionary Approaches To Minimizing Network Coding Resources This full text pper ws peer reviewed t the direction of IEEE Communictions Society suject mtter experts for puliction in the IEEE INFOCOM 2007 proceedings. Evolutionry Approches To Minimizing Network Coding

More information

2 Computing all Intersections of a Set of Segments Line Segment Intersection

2 Computing all Intersections of a Set of Segments Line Segment Intersection 15-451/651: Design & Anlysis of Algorithms Novemer 14, 2016 Lecture #21 Sweep-Line nd Segment Intersection lst chnged: Novemer 8, 2017 1 Preliminries The sweep-line prdigm is very powerful lgorithmic design

More information

Before We Begin. Introduction to Spatial Domain Filtering. Introduction to Digital Image Processing. Overview (1): Administrative Details (1):

Before We Begin. Introduction to Spatial Domain Filtering. Introduction to Digital Image Processing. Overview (1): Administrative Details (1): Overview (): Before We Begin Administrtive detils Review some questions to consider Winter 2006 Imge Enhncement in the Sptil Domin: Bsics of Sptil Filtering, Smoothing Sptil Filters, Order Sttistics Filters

More information

The Greedy Method. The Greedy Method

The Greedy Method. The Greedy Method Lists nd Itertors /8/26 Presenttion for use with the textook, Algorithm Design nd Applictions, y M. T. Goodrich nd R. Tmssi, Wiley, 25 The Greedy Method The Greedy Method The greedy method is generl lgorithm

More information

A Tautology Checker loosely related to Stålmarck s Algorithm by Martin Richards

A Tautology Checker loosely related to Stålmarck s Algorithm by Martin Richards A Tutology Checker loosely relted to Stålmrck s Algorithm y Mrtin Richrds mr@cl.cm.c.uk http://www.cl.cm.c.uk/users/mr/ University Computer Lortory New Museum Site Pemroke Street Cmridge, CB2 3QG Mrtin

More information

Fig.25: the Role of LEX

Fig.25: the Role of LEX The Lnguge for Specifying Lexicl Anlyzer We shll now study how to uild lexicl nlyzer from specifiction of tokens in the form of list of regulr expressions The discussion centers round the design of n existing

More information

COMP 423 lecture 11 Jan. 28, 2008

COMP 423 lecture 11 Jan. 28, 2008 COMP 423 lecture 11 Jn. 28, 2008 Up to now, we hve looked t how some symols in n lphet occur more frequently thn others nd how we cn sve its y using code such tht the codewords for more frequently occuring

More information

Mobile IP route optimization method for a carrier-scale IP network

Mobile IP route optimization method for a carrier-scale IP network Moile IP route optimiztion method for crrier-scle IP network Tkeshi Ihr, Hiroyuki Ohnishi, nd Ysushi Tkgi NTT Network Service Systems Lortories 3-9-11 Midori-cho, Musshino-shi, Tokyo 180-8585, Jpn Phone:

More information

In the last lecture, we discussed how valid tokens may be specified by regular expressions.

In the last lecture, we discussed how valid tokens may be specified by regular expressions. LECTURE 5 Scnning SYNTAX ANALYSIS We know from our previous lectures tht the process of verifying the syntx of the progrm is performed in two stges: Scnning: Identifying nd verifying tokens in progrm.

More information

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have Rndom Numers nd Monte Crlo Methods Rndom Numer Methods The integrtion methods discussed so fr ll re sed upon mking polynomil pproximtions to the integrnd. Another clss of numericl methods relies upon using

More information

Distributed Systems Principles and Paradigms

Distributed Systems Principles and Paradigms Distriuted Systems Principles nd Prdigms Chpter 11 (version April 7, 2008) Mrten vn Steen Vrije Universiteit Amsterdm, Fculty of Science Dept. Mthemtics nd Computer Science Room R4.20. Tel: (020) 598 7784

More information

CS143 Handout 07 Summer 2011 June 24 th, 2011 Written Set 1: Lexical Analysis

CS143 Handout 07 Summer 2011 June 24 th, 2011 Written Set 1: Lexical Analysis CS143 Hndout 07 Summer 2011 June 24 th, 2011 Written Set 1: Lexicl Anlysis In this first written ssignment, you'll get the chnce to ply round with the vrious constructions tht come up when doing lexicl

More information

Efficient Rerouting Algorithms for Congestion Mitigation

Efficient Rerouting Algorithms for Congestion Mitigation 2009 IEEE Computer Society Annul Symposium on VLSI Efficient Rerouting Algorithms for Congestion Mitigtion M. A. R. Chudhry*, Z. Asd, A. Sprintson, nd J. Hu Deprtment of Electricl nd Computer Engineering

More information

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li 2nd Interntionl Conference on Electronic & Mechnicl Engineering nd Informtion Technology (EMEIT-212) Complete Coverge Pth Plnning of Mobile Robot Bsed on Dynmic Progrmming Algorithm Peng Zhou, Zhong-min

More information

From Dependencies to Evaluation Strategies

From Dependencies to Evaluation Strategies From Dependencies to Evlution Strtegies Possile strtegies: 1 let the user define the evlution order 2 utomtic strtegy sed on the dependencies: use locl dependencies to determine which ttriutes to compute

More information

A dual of the rectangle-segmentation problem for binary matrices

A dual of the rectangle-segmentation problem for binary matrices A dul of the rectngle-segmenttion prolem for inry mtrices Thoms Klinowski Astrct We consider the prolem to decompose inry mtrix into smll numer of inry mtrices whose -entries form rectngle. We show tht

More information

On Achieving Optimal Throughput with Network Coding

On Achieving Optimal Throughput with Network Coding In IEEE INFOCOM On Achieving Optiml Throughput with Network Coding Zongpeng Li, Bochun Li, Dn Jing, Lp Chi Lu Astrct With the constrints of network topologies nd link cpcities, chieving the optiml end-to-end

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007 CS 88: Artificil Intelligence Fll 2007 Lecture : A* Serch 9/4/2007 Dn Klein UC Berkeley Mny slides over the course dpted from either Sturt Russell or Andrew Moore Announcements Sections: New section 06:

More information

Presentation Martin Randers

Presentation Martin Randers Presenttion Mrtin Rnders Outline Introduction Algorithms Implementtion nd experiments Memory consumption Summry Introduction Introduction Evolution of species cn e modelled in trees Trees consist of nodes

More information

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search.

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search. CS 88: Artificil Intelligence Fll 00 Lecture : A* Serch 9//00 A* Serch rph Serch Tody Heuristic Design Dn Klein UC Berkeley Multiple slides from Sturt Russell or Andrew Moore Recp: Serch Exmple: Pncke

More information

Slides for Data Mining by I. H. Witten and E. Frank

Slides for Data Mining by I. H. Witten and E. Frank Slides for Dt Mining y I. H. Witten nd E. Frnk Simplicity first Simple lgorithms often work very well! There re mny kinds of simple structure, eg: One ttriute does ll the work All ttriutes contriute eqully

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

If you are at the university, either physically or via the VPN, you can download the chapters of this book as PDFs.

If you are at the university, either physically or via the VPN, you can download the chapters of this book as PDFs. Lecture 5 Wlks, Trils, Pths nd Connectedness Reding: Some of the mteril in this lecture comes from Section 1.2 of Dieter Jungnickel (2008), Grphs, Networks nd Algorithms, 3rd edition, which is ville online

More information

10.5 Graphing Quadratic Functions

10.5 Graphing Quadratic Functions 0.5 Grphing Qudrtic Functions Now tht we cn solve qudrtic equtions, we wnt to lern how to grph the function ssocited with the qudrtic eqution. We cll this the qudrtic function. Grphs of Qudrtic Functions

More information

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5 CS321 Lnguges nd Compiler Design I Winter 2012 Lecture 5 1 FINITE AUTOMATA A non-deterministic finite utomton (NFA) consists of: An input lphet Σ, e.g. Σ =,. A set of sttes S, e.g. S = {1, 3, 5, 7, 11,

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

Unit #9 : Definite Integral Properties, Fundamental Theorem of Calculus

Unit #9 : Definite Integral Properties, Fundamental Theorem of Calculus Unit #9 : Definite Integrl Properties, Fundmentl Theorem of Clculus Gols: Identify properties of definite integrls Define odd nd even functions, nd reltionship to integrl vlues Introduce the Fundmentl

More information

What are suffix trees?

What are suffix trees? Suffix Trees 1 Wht re suffix trees? Allow lgorithm designers to store very lrge mount of informtion out strings while still keeping within liner spce Allow users to serch for new strings in the originl

More information

OUTPUT DELIVERY SYSTEM

OUTPUT DELIVERY SYSTEM Differences in ODS formtting for HTML with Proc Print nd Proc Report Lur L. M. Thornton, USDA-ARS, Animl Improvement Progrms Lortory, Beltsville, MD ABSTRACT While Proc Print is terrific tool for dt checking

More information

Efficient Algorithms For Optimizing Policy-Constrained Routing

Efficient Algorithms For Optimizing Policy-Constrained Routing Efficient Algorithms For Optimizing Policy-Constrined Routing Andrew R. Curtis curtis@cs.colostte.edu Ross M. McConnell rmm@cs.colostte.edu Dn Mssey mssey@cs.colostte.edu Astrct Routing policies ply n

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

Registering as a HPE Reseller. Quick Reference Guide for new Partners in Asia Pacific

Registering as a HPE Reseller. Quick Reference Guide for new Partners in Asia Pacific Registering s HPE Reseller Quick Reference Guide for new Prtners in Asi Pcific Registering s new Reseller prtner There re five min steps to e new Reseller prtner. Crete your Appliction Copyright 2017 Hewlett

More information

Approximation of Two-Dimensional Rectangle Packing

Approximation of Two-Dimensional Rectangle Packing pproximtion of Two-imensionl Rectngle Pcking Pinhong hen, Yn hen, Mudit Goel, Freddy Mng S70 Project Report, Spring 1999. My 18, 1999 1 Introduction 1-d in pcking nd -d in pcking re clssic NP-complete

More information

Network Interconnection: Bridging CS 571 Fall Kenneth L. Calvert All rights reserved

Network Interconnection: Bridging CS 571 Fall Kenneth L. Calvert All rights reserved Network Interconnection: Bridging CS 57 Fll 6 6 Kenneth L. Clvert All rights reserved The Prolem We know how to uild (rodcst) LANs Wnt to connect severl LANs together to overcome scling limits Recll: speed

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem Announcements Project : erch It s live! Due 9/. trt erly nd sk questions. It s longer thn most! Need prtner? Come up fter clss or try Pizz ections: cn go to ny, ut hve priority in your own C 88: Artificil

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016 Solving Prolems y Serching CS 486/686: Introduction to Artificil Intelligence Winter 2016 1 Introduction Serch ws one of the first topics studied in AI - Newell nd Simon (1961) Generl Prolem Solver Centrl

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method A New Lerning Algorithm for the MAXQ Hierrchicl Reinforcement Lerning Method Frzneh Mirzzdeh 1, Bbk Behsz 2, nd Hmid Beigy 1 1 Deprtment of Computer Engineering, Shrif University of Technology, Tehrn,

More information

GENERATING ORTHOIMAGES FOR CLOSE-RANGE OBJECTS BY AUTOMATICALLY DETECTING BREAKLINES

GENERATING ORTHOIMAGES FOR CLOSE-RANGE OBJECTS BY AUTOMATICALLY DETECTING BREAKLINES GENEATING OTHOIMAGES FO CLOSE-ANGE OBJECTS BY AUTOMATICALLY DETECTING BEAKLINES Efstrtios Stylinidis 1, Lzros Sechidis 1, Petros Ptis 1, Spiros Sptls 2 Aristotle University of Thessloniki 1 Deprtment of

More information

On network coding and routing in dynamic wireless multicast networks

On network coding and routing in dynamic wireless multicast networks On network coding nd routing in dynmic wireless multicst networks Trcey Ho, Ji-Qi Jin Cliforni Institute of Technology Psden, CA Emil: {tho, jin}@cltechedu Hrish Viswnthn Lucent Technologies, Bell Ls Murry

More information

9 Graph Cutting Procedures

9 Graph Cutting Procedures 9 Grph Cutting Procedures Lst clss we begn looking t how to embed rbitrry metrics into distributions of trees, nd proved the following theorem due to Brtl (1996): Theorem 9.1 (Brtl (1996)) Given metric

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

LECT-10, S-1 FP2P08, Javed I.

LECT-10, S-1 FP2P08, Javed I. A Course on Foundtions of Peer-to-Peer Systems & Applictions LECT-10, S-1 CS /799 Foundtion of Peer-to-Peer Applictions & Systems Kent Stte University Dept. of Computer Science www.cs.kent.edu/~jved/clss-p2p08

More information

Approximation by NURBS with free knots

Approximation by NURBS with free knots pproximtion by NURBS with free knots M Rndrinrivony G Brunnett echnicl University of Chemnitz Fculty of Computer Science Computer Grphics nd Visuliztion Strße der Ntionen 6 97 Chemnitz Germny Emil: mhrvo@informtiktu-chemnitzde

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

PARALLEL AND DISTRIBUTED COMPUTING

PARALLEL AND DISTRIBUTED COMPUTING PARALLEL AND DISTRIBUTED COMPUTING 2009/2010 1 st Semester Teste Jnury 9, 2010 Durtion: 2h00 - No extr mteril llowed. This includes notes, scrtch pper, clcultor, etc. - Give your nswers in the ville spce

More information

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork MA1008 Clculus nd Liner Algebr for Engineers Course Notes for Section B Stephen Wills Deprtment of Mthemtics University College Cork s.wills@ucc.ie http://euclid.ucc.ie/pges/stff/wills/teching/m1008/ma1008.html

More information

Transparent neutral-element elimination in MPI reduction operations

Transparent neutral-element elimination in MPI reduction operations Trnsprent neutrl-element elimintion in MPI reduction opertions Jesper Lrsson Träff Deprtment of Scientific Computing University of Vienn Disclimer Exploiting repetition nd sprsity in input for reducing

More information

EECS150 - Digital Design Lecture 23 - High-level Design and Optimization 3, Parallelism and Pipelining

EECS150 - Digital Design Lecture 23 - High-level Design and Optimization 3, Parallelism and Pipelining EECS150 - Digitl Design Lecture 23 - High-level Design nd Optimiztion 3, Prllelism nd Pipelining Nov 12, 2002 John Wwrzynek Fll 2002 EECS150 - Lec23-HL3 Pge 1 Prllelism Prllelism is the ct of doing more

More information

Some necessary and sufficient conditions for two variable orthogonal designs in order 44

Some necessary and sufficient conditions for two variable orthogonal designs in order 44 University of Wollongong Reserch Online Fculty of Informtics - Ppers (Archive) Fculty of Engineering n Informtion Sciences 1998 Some necessry n sufficient conitions for two vrile orthogonl esigns in orer

More information

12-B FRACTIONS AND DECIMALS

12-B FRACTIONS AND DECIMALS -B Frctions nd Decimls. () If ll four integers were negtive, their product would be positive, nd so could not equl one of them. If ll four integers were positive, their product would be much greter thn

More information

CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE

CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE CHAPTER III IMAGE DEWARPING (CALIBRATION) PROCEDURE 3.1 Scheimpflug Configurtion nd Perspective Distortion Scheimpflug criterion were found out to be the best lyout configurtion for Stereoscopic PIV, becuse

More information

Overview. Network characteristics. Network architecture. Data dissemination. Network characteristics (cont d) Mobile computing and databases

Overview. Network characteristics. Network architecture. Data dissemination. Network characteristics (cont d) Mobile computing and databases Overview Mobile computing nd dtbses Generl issues in mobile dt mngement Dt dissemintion Dt consistency Loction dependent queries Interfces Detils of brodcst disks thlis klfigopoulos Network rchitecture

More information

CSCI 3130: Formal Languages and Automata Theory Lecture 12 The Chinese University of Hong Kong, Fall 2011

CSCI 3130: Formal Languages and Automata Theory Lecture 12 The Chinese University of Hong Kong, Fall 2011 CSCI 3130: Forml Lnguges nd utomt Theory Lecture 12 The Chinese University of Hong Kong, Fll 2011 ndrej Bogdnov In progrmming lnguges, uilding prse trees is significnt tsk ecuse prse trees tell us the

More information

Registering as an HPE Reseller

Registering as an HPE Reseller Registering s n HPE Reseller Quick Reference Guide for new Prtners Mrch 2019 Registering s new Reseller prtner There re four min steps to register on the Prtner Redy Portl s new Reseller prtner: Appliction

More information

THE throughput of information transmission within a data

THE throughput of information transmission within a data IEEE TRANACTION ON INFORMATION THEORY A Constnt Bound on Throughput Improvement of Multicst Network Coding in Undirected Networks Zongpeng Li, Memer, IEEE Bochun Li, enior Memer, IEEE Lp Chi Lu Deprtment

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

An Efficient Divide and Conquer Algorithm for Exact Hazard Free Logic Minimization

An Efficient Divide and Conquer Algorithm for Exact Hazard Free Logic Minimization An Efficient Divide nd Conquer Algorithm for Exct Hzrd Free Logic Minimiztion J.W.J.M. Rutten, M.R.C.M. Berkelr, C.A.J. vn Eijk, M.A.J. Kolsteren Eindhoven University of Technology Informtion nd Communiction

More information

UT1553B BCRT True Dual-port Memory Interface

UT1553B BCRT True Dual-port Memory Interface UTMC APPICATION NOTE UT553B BCRT True Dul-port Memory Interfce INTRODUCTION The UTMC UT553B BCRT is monolithic CMOS integrted circuit tht provides comprehensive MI-STD- 553B Bus Controller nd Remote Terminl

More information

Intelligent Assembly/Disassembly System with a Haptic Device for Aircraft Parts Maintenance

Intelligent Assembly/Disassembly System with a Haptic Device for Aircraft Parts Maintenance Intelligent Assemly/Disssemly System with Hptic Device for Aircrft Prts Mintennce Christind nd Jungwon Yoon Mechnicl nd Aerospce Engineering nd ReCAPT, GyeongSng Ntionl University, Jinju, Gyeongnm 660-701,

More information

A Sparse Grid Representation for Dynamic Three-Dimensional Worlds

A Sparse Grid Representation for Dynamic Three-Dimensional Worlds A Sprse Grid Representtion for Dynmic Three-Dimensionl Worlds Nthn R. Sturtevnt Deprtment of Computer Science University of Denver Denver, CO, 80208 sturtevnt@cs.du.edu Astrct Grid representtions offer

More information

Algorithm Design (5) Text Search

Algorithm Design (5) Text Search Algorithm Design (5) Text Serch Tkshi Chikym School of Engineering The University of Tokyo Text Serch Find sustring tht mtches the given key string in text dt of lrge mount Key string: chr x[m] Text Dt:

More information

documents 1. Introduction

documents 1. Introduction www.ijcsi.org 4 Efficient structurl similrity computtion etween XML documents Ali Aïtelhdj Computer Science Deprtment, Fculty of Electricl Engineering nd Computer Science Mouloud Mmmeri University of Tizi-Ouzou

More information

MATH 25 CLASS 5 NOTES, SEP

MATH 25 CLASS 5 NOTES, SEP MATH 25 CLASS 5 NOTES, SEP 30 2011 Contents 1. A brief diversion: reltively prime numbers 1 2. Lest common multiples 3 3. Finding ll solutions to x + by = c 4 Quick links to definitions/theorems Euclid

More information

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES)

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) Numbers nd Opertions, Algebr, nd Functions 45. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) In sequence of terms involving eponentil growth, which the testing service lso clls geometric

More information

Performance enhancement of IEEE DCF using novel backoff algorithm

Performance enhancement of IEEE DCF using novel backoff algorithm Kuo et l. EURASIP Journl on Wireless Communictions nd Networking 212, 212:274 http://jis.eursipjournls.com/content/212/1/274 RESEARCH Open Access Performnce enhncement of IEEE 82.11 using novel ckoff lgorithm

More information

Rethinking Virtual Network Embedding: Substrate Support for Path Splitting and Migration

Rethinking Virtual Network Embedding: Substrate Support for Path Splitting and Migration Rethinking Virtul Network Emedding: Sustrte Support for Pth Splitting nd Migrtion Minln Yu, Yung Yi, Jennifer Rexford, Mung Ching Computer Science, Princeton University, Emil: {minlnyu,jrex}@cs.princeton.edu

More information

Efficient K-NN Search in Polyphonic Music Databases Using a Lower Bounding Mechanism

Efficient K-NN Search in Polyphonic Music Databases Using a Lower Bounding Mechanism Efficient K-NN Serch in Polyphonic Music Dtses Using Lower Bounding Mechnism Ning-Hn Liu Deprtment of Computer Science Ntionl Tsing Hu University Hsinchu,Tiwn 300, R.O.C 886-3-575679 nhliou@yhoo.com.tw

More information

View, evaluate, and publish assignments using the Assignment dropbox.

View, evaluate, and publish assignments using the Assignment dropbox. Blckord Lerning System CE 6 Mnging Assignments Competencies After reding this document, you will e le to: Crete ssignments using the Assignment tool. View, evlute, nd pulish ssignments using the Assignment

More information

Engineer To Engineer Note

Engineer To Engineer Note Engineer To Engineer Note EE-186 Technicl Notes on using Anlog Devices' DSP components nd development tools Contct our technicl support by phone: (800) ANALOG-D or e-mil: dsp.support@nlog.com Or visit

More information

Shortcomings with Tree-Structured Edge Encodings for Neural Networks

Shortcomings with Tree-Structured Edge Encodings for Neural Networks Shortcomings with Tree-Structured Edge Encodings for Neurl Networks Gregory S. Horny QSS Group Inc., NASA Ames Reserch Center Mil Stop 269-3, Moffett Field, CA 94035-1000 horny@emil.rc.ns.gov Astrct. In

More information

Graphs with at most two trees in a forest building process

Graphs with at most two trees in a forest building process Grphs with t most two trees in forest uilding process rxiv:802.0533v [mth.co] 4 Fe 208 Steve Butler Mis Hmnk Mrie Hrdt Astrct Given grph, we cn form spnning forest y first sorting the edges in some order,

More information

A Heuristic Approach for Discovering Reference Models by Mining Process Model Variants

A Heuristic Approach for Discovering Reference Models by Mining Process Model Variants A Heuristic Approch for Discovering Reference Models by Mining Process Model Vrints Chen Li 1, Mnfred Reichert 2, nd Andres Wombcher 3 1 Informtion System Group, University of Twente, The Netherlnds lic@cs.utwente.nl

More information

George Boole. IT 3123 Hardware and Software Concepts. Switching Algebra. Boolean Functions. Boolean Functions. Truth Tables

George Boole. IT 3123 Hardware and Software Concepts. Switching Algebra. Boolean Functions. Boolean Functions. Truth Tables George Boole IT 3123 Hrdwre nd Softwre Concepts My 28 Digitl Logic The Little Mn Computer 1815 1864 British mthemticin nd philosopher Mny contriutions to mthemtics. Boolen lger: n lger over finite sets

More information

LR Parsing, Part 2. Constructing Parse Tables. Need to Automatically Construct LR Parse Tables: Action and GOTO Table

LR Parsing, Part 2. Constructing Parse Tables. Need to Automatically Construct LR Parse Tables: Action and GOTO Table TDDD55 Compilers nd Interpreters TDDB44 Compiler Construction LR Prsing, Prt 2 Constructing Prse Tles Prse tle construction Grmmr conflict hndling Ctegories of LR Grmmrs nd Prsers Peter Fritzson, Christoph

More information

Position Heaps: A Simple and Dynamic Text Indexing Data Structure

Position Heaps: A Simple and Dynamic Text Indexing Data Structure Position Heps: A Simple nd Dynmic Text Indexing Dt Structure Andrzej Ehrenfeucht, Ross M. McConnell, Niss Osheim, Sung-Whn Woo Dept. of Computer Science, 40 UCB, University of Colordo t Boulder, Boulder,

More information

Simrad ES80. Software Release Note Introduction

Simrad ES80. Software Release Note Introduction Simrd ES80 Softwre Relese 1.3.0 Introduction This document descries the chnges introduced with the new softwre version. Product: ES80 Softwre version: 1.3.0 This softwre controls ll functionlity in the

More information

Pointwise convergence need not behave well with respect to standard properties such as continuity.

Pointwise convergence need not behave well with respect to standard properties such as continuity. Chpter 3 Uniform Convergence Lecture 9 Sequences of functions re of gret importnce in mny res of pure nd pplied mthemtics, nd their properties cn often be studied in the context of metric spces, s in Exmples

More information

Video-rate Image Segmentation by means of Region Splitting and Merging

Video-rate Image Segmentation by means of Region Splitting and Merging Video-rte Imge Segmenttion y mens of Region Splitting nd Merging Knur Anej, Florence Lguzet, Lionel Lcssgne, Alin Merigot Institute for Fundmentl Electronics, University of Pris South Orsy, Frnce knur.nej@gmil.com,

More information

Ma/CS 6b Class 1: Graph Recap

Ma/CS 6b Class 1: Graph Recap M/CS 6 Clss 1: Grph Recp By Adm Sheffer Course Detils Adm Sheffer. Office hour: Tuesdys 4pm. dmsh@cltech.edu TA: Victor Kstkin. Office hour: Tuesdys 7pm. 1:00 Mondy, Wednesdy, nd Fridy. http://www.mth.cltech.edu/~2014-15/2term/m006/

More information

SUPPLEMENTARY INFORMATION

SUPPLEMENTARY INFORMATION Supplementry Figure y (m) x (m) prllel perpendiculr Distnce (m) Bird Stndrd devition for distnce (m) c 6 prllel perpendiculr 4 doi:.8/nture99 SUPPLEMENTARY FIGURE Confirmtion tht movement within the flock

More information

Computer Networks 53 (2009) Contents lists available at ScienceDirect. Computer Networks. journal homepage:

Computer Networks 53 (2009) Contents lists available at ScienceDirect. Computer Networks. journal homepage: Computer Networks 53 (29) 252 2516 Contents lists ville t ScienceDirect Computer Networks journl homepge: www.elsevier.com/locte/comnet Congestion-wre chnnel ssignment for multi-chnnel wireless mesh networks

More information

CSCI 104. Rafael Ferreira da Silva. Slides adapted from: Mark Redekopp and David Kempe

CSCI 104. Rafael Ferreira da Silva. Slides adapted from: Mark Redekopp and David Kempe CSCI 0 fel Ferreir d Silv rfsilv@isi.edu Slides dpted from: Mrk edekopp nd Dvid Kempe LOG STUCTUED MEGE TEES Series Summtion eview Let n = + + + + k $ = #%& #. Wht is n? n = k+ - Wht is log () + log ()

More information

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Solving Prolems y Serching CS 486/686: Introduction to Artificil Intelligence 1 Introduction Serch ws one of the first topics studied in AI - Newell nd Simon (1961) Generl Prolem Solver Centrl component

More information

Midterm 2 Sample solution

Midterm 2 Sample solution Nme: Instructions Midterm 2 Smple solution CMSC 430 Introduction to Compilers Fll 2012 November 28, 2012 This exm contins 9 pges, including this one. Mke sure you hve ll the pges. Write your nme on the

More information

Notes for Graph Theory

Notes for Graph Theory Notes for Grph Theory These re notes I wrote up for my grph theory clss in 06. They contin most of the topics typiclly found in grph theory course. There re proofs of lot of the results, ut not of everything.

More information

a(e, x) = x. Diagrammatically, this is encoded as the following commutative diagrams / X

a(e, x) = x. Diagrammatically, this is encoded as the following commutative diagrams / X 4. Mon, Sept. 30 Lst time, we defined the quotient topology coming from continuous surjection q : X! Y. Recll tht q is quotient mp (nd Y hs the quotient topology) if V Y is open precisely when q (V ) X

More information

Inference of node replacement graph grammars

Inference of node replacement graph grammars Glley Proof 22/6/27; :6 File: id293.tex; BOKCTP/Hin p. Intelligent Dt Anlysis (27) 24 IOS Press Inference of node replcement grph grmmrs Jcek P. Kukluk, Lwrence B. Holder nd Dine J. Cook Deprtment of Computer

More information

On Computation and Resource Management in Networked Embedded Systems

On Computation and Resource Management in Networked Embedded Systems On Computtion nd Resource Mngement in Networed Embedded Systems Soheil Ghisi Krlene Nguyen Elheh Bozorgzdeh Mjid Srrfzdeh Computer Science Deprtment University of Cliforni, Los Angeles, CA 90095 soheil,

More information

Digital Signal Processing: A Hardware-Based Approach

Digital Signal Processing: A Hardware-Based Approach Digitl Signl Processing: A Hrdwre-Bsed Approch Roert Esposito Electricl nd Computer Engineering Temple University troduction Teching Digitl Signl Processing (DSP) hs included the utilition of simultion

More information

this grammar generates the following language: Because this symbol will also be used in a later step, it receives the

this grammar generates the following language: Because this symbol will also be used in a later step, it receives the LR() nlysis Drwcks of LR(). Look-hed symols s eplined efore, concerning LR(), it is possile to consult the net set to determine, in the reduction sttes, for which symols it would e possile to perform reductions.

More information

Agilent Mass Hunter Software

Agilent Mass Hunter Software Agilent Mss Hunter Softwre Quick Strt Guide Use this guide to get strted with the Mss Hunter softwre. Wht is Mss Hunter Softwre? Mss Hunter is n integrl prt of Agilent TOF softwre (version A.02.00). Mss

More information

INTRODUCTION TO SIMPLICIAL COMPLEXES

INTRODUCTION TO SIMPLICIAL COMPLEXES INTRODUCTION TO SIMPLICIAL COMPLEXES CASEY KELLEHER AND ALESSANDRA PANTANO 0.1. Introduction. In this ctivity set we re going to introduce notion from Algebric Topology clled simplicil homology. The min

More information

Section 3.1: Sequences and Series

Section 3.1: Sequences and Series Section.: Sequences d Series Sequences Let s strt out with the definition of sequence: sequence: ordered list of numbers, often with definite pttern Recll tht in set, order doesn t mtter so this is one

More information

CS481: Bioinformatics Algorithms

CS481: Bioinformatics Algorithms CS481: Bioinformtics Algorithms Cn Alkn EA509 clkn@cs.ilkent.edu.tr http://www.cs.ilkent.edu.tr/~clkn/teching/cs481/ EXACT STRING MATCHING Fingerprint ide Assume: We cn compute fingerprint f(p) of P in

More information

A modified directed search domain algorithm for multiobjective engineering and design optimization

A modified directed search domain algorithm for multiobjective engineering and design optimization Struct Multidisc Optim DOI.7/s58-3-946- RESEARCH PAPER A modified directed serch domin lgorithm for multiojective engineering nd design optimiztion Tohid Erfni Sergey V. Utyuzhnikov Brin Kolo Received:

More information

TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS

TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS TOWARDS GRADIENT BASED AERODYNAMIC OPTIMIZATION OF WIND TURBINE BLADES USING OVERSET GRIDS S. H. Jongsm E. T. A. vn de Weide H. W. M. Hoeijmkers Overset symposium 10-18-2012 Deprtment of mechnicl engineering

More information