New User Special Price Expires in

Let's log you in.

Sign in with Facebook


Don't have a StudySoup account? Create one here!


Create a StudySoup account

Be part of our community, it's free to join!

Sign up with Facebook


Create your account
By creating an account you agree to StudySoup's terms and conditions and privacy policy

Already have a StudySoup account? Login here

Research Seminar

by: Mireya Heidenreich

Research Seminar CS 7123

Mireya Heidenreich
GPA 3.55


Almost Ready


These notes were just uploaded, and will be ready to view shortly.

Purchase these notes here, or revisit this page.

Either way, we'll remind you when they're ready :)

Preview These Notes for FREE

Get a free preview of these Notes, just enter your email below.

Unlock Preview
Unlock Preview

Preview these materials now for free

Why put in your email? Get access to more of this material and other relevant free materials for your school

View Preview

About this Document

Class Notes
25 ?




Popular in Course

Popular in ComputerScienence

This 161 page Class Notes was uploaded by Mireya Heidenreich on Thursday October 29, 2015. The Class Notes belongs to CS 7123 at University of Texas at San Antonio taught by Staff in Fall. Since its upload, it has received 12 views. For similar materials see /class/231397/cs-7123-university-of-texas-at-san-antonio in ComputerScienence at University of Texas at San Antonio.

Similar to CS 7123 at UTSA

Popular in ComputerScienence


Reviews for Research Seminar


Report this Material


What is Karma?


Karma is the currency of StudySoup.

You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 10/29/15
Towards Justintime Middleware Architecture Charles Zhang Dapeng Gao and HansArno Jacobson AOSD 2005 Presented by Humayun Zafar Outline Problem Statement Preamble Middleware Architectures JustInTime Customization JiM Dependencies Constraints and Convolution Descriptions The Abacus Arachne Related Work Conclusion Problem Statement 0 Prepostulated architectures PPA o Middleware functionality independently conceived and packaged as framework libraries according to which applications are developed 0 They have limitations Impedance mismatch Application obliviousness Domain limitations Excessive configurability Impedance Mismatch 0 Existing middleware architectures designed according to specific application domains o Enterprise systems safety critical systems et cetera 0 User applications might not be categorized in the same way Functionally redundant over fit or insufficient under fit or incomplete partial fit Application Obliviousness o PPA s do not take full advantage of information embodied in user applications regarding required middleware features o Data types and invocations styles 0 Difficult to predesign a customizable and adaptive system which takes these into consideration Domain Limitations 0 Solutions are domain specific 0 gt Problematic in other domains 0 Due to nonmodularity 0 Again difficulty lies in all the variations in each and every user application scenario Excessive Configurability 0 Software configuration independent of middleware architecture o Left either minimal in terms of application specific or primarily as user s responsibility 0 A daunting task o Configuration possibilities grow very fast 0 ldea Paradigm shift in the construction of middleware o Postpostulation o Use knowledge embedded in the user application Preamble 0 AOSD o Methodology that allows for the explicit expression of concerns that 0 Can be reused in other designs though flexible composition capabilities 0 Many technologies exist HyperJ AspectJ Lasagne o Most are Java oriented Middleware Architectures o What is it Set of services which facilitate development of distributed systems Narrowed scope in this paper 0 Software substrate which enables transparent remote invocations of services 0 CORBA DCOM and Net 0 Types RPC Client makes calls to procedures running on remote systems Can be synchronous or asynchronous Message Oriented Middleware Messages sent to the client are collected and stored until they are acted upon while the client continues with other processing ORB Allows for applications to send objects and request services in an objectoriented system SQLOriented Data Access middleware between applications and database servers Middleware Architectures CORBA 0 What is CORBA 0 Common Object Request Broker Architecture 0 Standard architecture for distributed object systems 0 OMG Object Management Group 0 CORBA Services 0 Standard CORBA objects with IDL interfaces sometimes referred to as Object Services 0 Naming service 0 IDL Interfaces 0 Support the specification of object interfaces 0 An object interface indicates the operations the object supports but not how they are implemented IDL Example module StockObjects struct Quote string symbol long attime double price long volume exception Unknown interface Stock Returns the current stock quote Quote getquote raisesUnknown Sets the current stock quote void setquotein Quote stockquote Provides the stock description eg company name readonly attribute string description IDL to Java Binding Constructs Module Package Namespace Interface Interface Abstract class Operation Method Member function Attribute Pair of methods Pair of functions Excep on Excep on Excep on IDL to Java Binding Data types 3 Boolean Boolean Charwchar Char Octet Byte Shortunsigned short Short Longunsigned long lnt Long longunsigned long long long float float Double double Stringwstring String JustlnTime Customization 5r 0 The program is obsolete by the time it is done Example Kodak and Dale Labs a Rationale Distribution intentions of the user applications drive the final composition of middleware o Contrary to traditional construction methods JiM JustIn Time Middleware 0 Metaphorically borrowed from JIT compilation in Java 0 Executable middleware generated at an appropriate time o Assumptions o Paradigm The middleware core 0 Vertical and horizontal architecture o Limited to RPC based middleware o Most effective in a targeted distributed services environment JiM Stages a Four highlevel stages a Process bootstrapped by explicit acquisition Suggestions Initial Manifest v D Feamre 4 Functional Inference 9 Dependency HUIES inferredMani39i esr 39 lt Verification 4 9 Composition Constrai nts Fina Manifest Synthesis lt Q Convolution Descriptions Mindaware implementarion Dependencies constraints 33 and convolution descriptions o Three specifications in addition to the middleware implementation o Feature dependency Implementation of certain functionality is composed from other functionalities 0 Example String data type results in String and char being included in the feature manifest o Composition constraints Dictate inclusions and exclusions of functionalities reflecting certain conditions o Example A J2ME application requires omission of certain functionalities due to lack of support of J2ME VM o Convolution descriptions Propositions describing the interactions among features Will revisit this Functionality Acquisition 0 Goal Obtain minimum set of middleware functionalities as desired by the user application 0 Two stages a Explicit acquisition Direct collection of functionalities o Inference Adjustment according to functional dependencies Explicit Acquisition 0 Two basic forms a Automatic feature extraction Examine the program source and detect middleware functionalities Example CORBA DCOM IDL o User preference indication User selects additional features in foreseeing their future uses Inference 0 Reason about functional dependency rules and possibly include additional functionalities 0 Process must provide two guarantees a Each functionality in the manifest must have its functionality dependency satisfied a The inferred manifest must be a minimum superset of the explicit manifest 0 An appropriate cost model needs to be employed by inference algorithms to gauge what the minimum should be Verification 39 o Inferred manifest is checked against composition constraints a Violations of these constraints can be used to o Prune choices of feature supersets Reinitiate the acquisition process Target Synthesis 0 Tailored middleware is composed in strict accordance with the final manifest generated at the inference stage 0 Organize code space of middleware features via 0 Structural schema o Implementation selection Structural Schema 0 Goal Make the code space of middleware features comprehensive to the synthesizer 0 Traditional structure of the code space is based on name spaces 0 Structural schema for JiM must make the following properties explicit for each aspect o Differentiation between implementation and interaction and the set of binary relationships with other aspects Implementation selection 0 Primary role of the JiM synthesizer 0 According to the final manifest determines and includes the correct constituents of a feature implementation in the final middleware implementation 0 Output 0 Build configuration 0 Executable middleware instance Functional Verification 339 0 Typically verified through integration tests in which individual tests are provided for all the packaged functionalities o JiIVI o Such tests difficult to devise a Advantage High degree of orthogonality o Synthesizer s job to automatically assemble the composite test plan a Can be integrated with the synthesizer if structural schema is used Abacus 3 0 Prototype implementation of JiM 0 Architecture based on a very high level of configurability along two dimensions o Vertical dimension preserves the original hierarchical configurability o Horizontal dimension modularizes crosscutting concerns Vertical Configurability 0 Changing parts of the hierarchical structure 0 Abacus can be vertically configured System property tables Java equivalence of the environment variable mechanism 0 Policies CORBA specific way of fine tuning the functionalities of ORB Horizontal Configurability o Vast amounts of aspects are available a Includes lDL language data types Categories Features Scn cr Data Integer String Double Typos Fixed Flo4t Long VVitlL Chairlift01 Vhlc String Remote Invocation SLninntiCs Synchrony Asyucln39ony Oncway P ussing by Vi 11n C39OR BA h 39ush uctin39cs Policies Impchmutation Repository lntci39facc Repository Type Codc Collocation OptimizzltionIntcrccptors Caching Fault Tolerance Ol jcct Disposal Romotc invocation Style D11 D51 Any Dniuinjc Any Rogionul Support Lo calc Enco cling c0 nvcrsion Arachne o The Abacus synthesizer 0 Properties o Stages of JIT customization in Abacus are integrated and automated o Composed of three components o Aspectaware IDL compiler o Java source parser o Inference engine Aspectaware IDL compiler 5r 0 Carries out two key functions a Produces modular skeletons and stubs o Explicitly acquires middleware functionalities from service declarations o Compiler collects IDL language elements that can be used to initiate functionality inference Dependencies Constraints and Convolution Descriptions 0 Dependencies in Arachne specified by horizontal 1 aspect aspect1 aspect2 dependenCy defmltlon g mOdUzlgpaeEi aspeth aspectS aspect4 language HDL 4 5 0 Figure 0 Identifiers of aspects declared using the aspect keyword line 1 o Dependency rules of the same nature grouped and scoped with a name line 2 Aspect1 depends on aspect2 together with either aspect3 or aspect4 Dependencies of Horizontal Features in Abacus 0 Exception made for the dependency rule core 0 Activated by default 0 Specifies the minimum valid configuration of Abacus o A valid Abacus instance must support either synchronous or asynchronous communication aspect any dynamic pi synchrony asynchrony Core core synchrony asynchrony any int string ulonglong dii dynamic dsi dynamic pi dynamic dynamic any ulonglong longlong IR DII DSItypecode convert locale wstring wchar Constraint Specifications in Abacus 0 Constraints are used to limit the freedom of feature selection o Denies the selection of certain features in certain environments 0 Left side of Boolean expression is the name on the constraint 0 Figure 0 Constraints for both J2ME and desktop platforms o J2ME constraints are mostly reflections of platform incompatibilities between JZSE and J2ME 0 0 O O 0 O j2me resource valuetype finalize precision double fixed infrastructure l applet desktop reflection dii dsi typeinfo IR IMR Descriptions of Constraints 0 Division into separate constraining rules is to provide more specific quot and user UUL CtiOll Mich 15 quotdoublequot and quot xmlquot not in JZAIE typo info liy typo infur11mtiltgtn are Conunul y used in upon violations 0 Desktop constraints are illustrative 0 Figure 0 Descriptions of the constraints Convolution Descriptions in Abacus 0 Describe the oneto many relationships between any two horizontal features 0 Figure 0 Left side of Boolean expression is a selectable feature o Helpful to the final synthesis convolutiondescriptions double any valuetype finalize DII fixed any IR IR policy PI policy IMR oneway PI ulonglong an valuetype any IR policy PI valuetype dii any dynamic IR longlong double any valuetype oneway collocation PI DII DSI IR wchar any convert valuetype Inference Algorithm in Abacus 5 o Responsible for selecting unspecified features 0 Satisfies all functional dependency rules o If multiple choices for selections exist the algorithm guarantees a minimum superset of the initial subset o Terminology 0 Feature state Can be in one of three states 0 Selected true excluded false and unspecified unknown o Rule Activation A dependency rule r is activated if the feature on the left side of r is selected More Terminology 0 Minterm This algorithm only deals with rules written in the form of sumof products 0 Cost function Associates each feature with a positive number representing the cost of including this feature The Algorithm Itself Require initial manifest F icpeniicucy rules I o All rules requiring one minterm evaluated to unknown 0 Iterates each r and invokes the evaluate function on r against the inferred manifest 0 Case unknown 0 New features are added to the manifest to satisfy r Let F F loop for each 139 E R 0 if I is activated Irv F I then Boolean result evaluatcir F39 if result true then GOTO next 139 else if result false then RETURN error else for each imkown f in I39 do Hot f to l or i so that I evaluates to 1 end for end if end if end for if F F then BREAK else F F 39 end if end loop Set all iu mowu features in F to false RETURN F Packaging Schema in Abacus 5 0 Code of an aspect needs to be packaged according to a welldefined schema o Relevant parts of the aspect can be correctly selected for the final synthesis 0 Figure o Implementation and interaction of an aspect are separated into two sub trees funct imp aspeci name interaction aspectname ltlt more aspect names gtgt default rest asgedname ltlt more aspect names gtgt ltlt regular name space gtgt Synthesis Process 3 0 Guided by the Arachne user interface 0 Initiated by loading IDL definitions and application source locations 0 When should an aspect be activated 0 During the source analysis when linking the class types in the application source to the implementation class types of an aspect o Activated features are marked on the feature pallet on which user can make further selections and adjustments More Synthesis 0 Complete set of explicitly acquired features is processed by the inference engine a gt generates the final manifest of features 0 With help of the packaging schema and convolution descriptions the end result is Final Synthesis Procedure Require nal umuifest F Convolution lcscrition set C39 O O Utput for all f in F do if Ely E C39 such that f activates r then for Cfl C l f39 011 the right HillL of 39 do f Include all modules in the path flIIr39fiunuliInplmm nfnHon Include all lllm llllcs in the path quotf 2 quotm439 lllCllulu all 1110lulcs 1n the path f 9 uh I mfin gt f quot end for 0 Feature end if end for 0 Easy to add a new aspect feature if it conforms to the code naming schema Some Hand Waving o Integration test synthesis 0 Functionality of the synthesized Arachne instance can be verified through a synthesized set of test cases 0 Abacus Test cases written in JUnit independently for each feature Schema for testcase organization is the same as for the features and the same synthesis procedure is used So when a particular feature is selected its test cases are also woven into the final test plan Related Work 55quot o Customizable Middleware o MicroQoSCorba McKinnon et al IEEE 2003 project o Targets resourceconstrained environments and makes automatic adjustments o Uses IDL compilers to collect required middleware features 0 Unlike Abacus source code analysis is missing 0 Product Lines and Generative Programming 0 GenVoca Batory et al IEEE 2004 0 Synthesis of complex systems from incrementally adding features to simple ones Conclusion 3 o Prepostulated middleware poses challenges o Which features are needed and when 0 In the case of middleware architecture a Best presumption is not to make any presumptions at all 0 This paper proposed a delaying of the composition of the middleware architecture till after the user application is designed Amost Over 0 Proposed JiM as having multistage processes a Impedance mismatch 0 Questions Please say no Quantum Computation Presented by Samira Khan Outline Introduction I Qubit I Gates Algorithm Quantum Cryptography Realization Conclusion Quantum Computation Introduction Moore s Law number of transistors on a chip doubles about every two years I Integrated circuits are becoming smaller and smaller I Soon quantum effects will begin to interfere why not build a quantum computer Quantum Computation Introduction Law mm M Hue products Ddayad 45ml 32 nm Mum Capx Mmzm Much Mimi PERCMD More s mum mm a 1m immmns munMl W qu w mm naemp v 0mm MusroAscwmuswcmmm Mm wumuouva L w Mm I NW NWUAHHN MW Murmnvnm mm mm mm v mm 1 mum 4m MMVJN I I vm39mn hardanzhunapuu m nynptu a 525mmquot W wqummk WWW 75 huhhunrele L EHHIH mu m u m Quanhm Campuzman Introduction IIIII upticul lithography ll 1 I1 39 I III dium er of hydrogen atom I 39 I I I I I I I I I I I I lm I950 I9 3111 fear Jmmuinr size micrnns Quantum Computation I Introduction How small can they be 1 0396 meters 10 meters 1 0quot meters TODAY TOMORROW Here Quantum mechanics comes into play Quantum Computation Quantum bit Qubit Mathematical Concept Classical bit has state 0 or 1 Qubit can also have state IO or state 1 But is also possible to be in linear combination of states called superposition wgta0gt llgt u Alpha and beta are complex numbers Quantum Computation Qubit wgta0gt llgt When we measure a qubit We will get 0 with probabily oc2 and 1 with probability IBI2 oc2 B2 1 a 12 IO I1 means that if we measure this qubit We will get state IO 50 times and I state 1 50 of the time Quantum Computation The cat paradox Schrodinger39s Cat 1 A cat is placed in a box 1 together with a radioactive atom a If the atom decays and the geiger counter detects an alpha particle the hammer hits a flask of prussic acid HCN killing the cat a Is the cat dead or alive Quantum Computation 9 The cat paradox Fquot39BWW 139 w Quantum Computation 10 I The cat paradox l The answer according to quantum mechanics is that it is 50 dead and 50 alive the cat39s fate is tied to the wave function of the atom which is itself in a superposition of decayed and undecayed states he cat must itself be in a superposition of dead and alive states before the 0 server opens the box observesquot the cat and collapsesquot it39s wave function V K Qumtum Computation Qubit GOOD NEWS quantum parallel processing on 2N inputs Cowo Example N3 qubits P a0000 a1I001 a2 I010a3 O11gt 054 100gt a5l101a6 110gt a7111 BAD NEWS Measurement gives random result eg w 1o1gt Quantum Computation Qubit In quantum mechanics the Bloch sphere is a geometrical representation of a 2level quantum system it is the pure state space of a 1 gubit quantum register wgta0gt 1gt Can be written as Iwgtco ogtwtagtsirg 1gt Quantum Computation Qubit Qumtum Computmon Qubit Realization of qubit a Nuclear spin positive negative u Polarization of a photon I let I right a Electron state ground I excited Quantum Computation I Realization of qubit Spin W C05lt92IDIgt em siN021gt Spin12 pariicle 390 H LEM5 U HUM5 m MW IT JDvE Bloch sphere Qnmmm c mum Realization of qubit polarization W COS02Ogt mime2m Polarization of a photon um HunVi WWW 7 WW Mi 7 mm M IR zxLgtgtNi WWW Hgt WNWVi IV Poincare sphere Qumtum Computmon 17 Realization of qubit atorn state w COS02IO Mama2m gt 11gt Twolevel atom q 2 lo 9 0 9 EDA5 09 JEDVi 9 739Egt5 l9 FDNE Bloch sphere Quantum Curnpuuhm Single Qubit gates Constraint on gates UTU I Gate can be presented as matrix Quantum NOT gate X 01 10 XW 8 a 0gt gtX gt1gt 1gt gtX gt0gt 060gt 1gt gt 0gt061gt Quantum Computation 19 Other Single Qubit gates Pauli Z gate a This gate flips the sign of 1 to give 1 Z 0 1 0gt gtZ gt0gt 1gt gtZ gt 1gt alO81gt gtZ gta0gt B1gt Quantum Computation 20 Other Single Qubit gates Hadamard gate a This is also called square root of NOT gate H1 1 1 51 1 0gt gtH gt0gt1gt 1gt gtH gt0H1gt 1 1 a0gt 1gt gtH gta l 0gt1gt l 0gt I1gt Quantum Computation 21 Multi Qubit Gate ControlledNOT gate IX 6 IV IX I xgtelygt Quantum Computation 22 Multi Qubit Gate CNOT OCH H000 O O 1 O OOHO 0 00gt gt CNOT gt 00gt 01gt gt CNOT gt 01gt 10gt gt CNOT a 11gt 11gt gt CNOT gt 10gt Quantum Computation 23 Quantum Gate Array I 1 bit Full Adder ICgt I J O gt ICgt Xgt j J O Xgt lygt J T gt lygt 0gt sgt 0gt 0gt IC gt 1gt Let Icgt 1gt Xgt 0gt ygt 1gt sgt 0gt lc gt 1gt Quantum Computation 24 Shor s Algorithm factorizing integer into primes Input a composite integer N Output a nontrivial factor of N Runtime Olog N3 operations Quantum Computation 25 I Shor s Algorithm he mummy ew xewice in his umpell mu nmmanity Shor39s historic factoring algorithm demonstrated with IBM39s testtube quantum computer San Jose 19 December 2001 Scleant S at IBM39s 3 e w a E a Q s 1a a lt n E e 2 a a s m s 3 n L w a 392 m in the latest Issue of the setentme journal Nature a team of IBM seienttsts and Stanford Universtty graduate students report the rst demonstra Ion o s on ntn method devetonen In 1994 bv ATampT setentust Peter snot tot Quantum Computation 26 Grover s Algorithm The database search problem Given an unsorted database containing n items but only one marked item find the address of the marked item with a minimum number of database calls Grover s algorithm uses On calls Grover39s Search 0 1 O 0 X1 X2 X3 X I Find i for which X is 1 fl Quantum Computation 27 Quantum Cryptography Quantum Codebreaking Shor s algorithm Factoring is easy with a quantum computer Quantum computing can efficiently break El RSA 1 Discrete logarithm problem DiffieHellman key exchange a Ellipticcurve cryptographic systems If a quantum computer is ever built Cryptography will fall apart Quantum Computation 28 Quantum Key Distribution Absolute security based on fundamental laws of quantum mechanics Allow two persons who share a small amount of authentication information to communicate in absolute security in the presence of an eavesdropper Any eavesdropping attack will essentially always be Quantum Computation 29 Quantum Key Distribution Absolute security based on fundamental laws of quantum mechanics Allow two persons who share a small amount of authentication information to communicate in absolute security in the presence of an eavesdropper Any eavesdropping attack will essentially always be Quantum Computation 30 Physical Implementation Ion trap Neutral atoms in trap Cavity QED NMR Quantum Dots And many more Quantum Computation 31 Physical Implementation Michigan Ion Trap machine Quantmn Computation 32 Physical Implementation Japanese NMR machine Quantum Computation 33 Where are we now DWave System Inc a Launched orion world s first commercial 16 bit quantum computer on February 13th 2007 a DWave demonstrated three different applications at the Computer History Museum in Mountain View California Pattern matching performed a search for a similar compound to a known drug within a database of molecules computed a seating arrangement for an event subject to compatibilities and incompatibilities between guests I SUdOkU Quantum Computation 34 DWave System Inc Quantum Computation 35 J D Wave System Inc my um Quantum Compumon Selected Sources I Oxford s Centre for Quantum Computation wwwgubitorg httpwwwquantikiorg Quantum Emnputation and Quantum information Quantum Computation and Quantum Information by Michael A Nielsen Author Isaac L Chuang Author Quantum Computation 37 Conclusion We will definitely have quantum computers in our home in the future Cryptography has to adopt the quantum computation approach Many physical limitations Ways to go Quantum Computation 38 do n 35quot A n E i x f 7 l 14 1x V 1 I L I x I i Qiu et a1 NAR 2005 Presented by Kihoon Yoon Cl 0 O 0 Background 0 Transcription amp Translation 0 microRNA vs siRNA Why RNAi Purpose amp Problems in vitro and in silico Proposed Solution Possible improvements Background Translation amp Transcription W 0 A gene is transcribed to a llmTranscnplwn mRNA then mRNA is RNA translated to a protein Jm Postltansc g tj Thus possible gene quotm expression regulation points 6 emne are before transcription pre Winsome I mm transcription regulation Q transcription control T39m39a m Polypephde between transcriptlon and d Postmmm translat10npost momma Upon Itself transcription regulation translational control 39 a wmwewvc Active Protein Background microRNA 0 microRNA certain parts of the genome are transcribed into short RNA molecules that fold back on themselves in a hairpin shape to create a double strand primary miRNA structure primiRNA The Dicer enzyme then cuts 20 25 nucleotides from the base of the hairpin to release the mature miRNA Animation httpWWWnaturecomngsupplementsmicrornasVide ohtml Background RNAi RNA interference RNAi 0 A mechanism in the cell biology of many eukaryotes in which fragments of double stranded ribonucleic acid dsRNA interfere with the expression of a particular gene Whose sequence is complementary to the dsRNA Animations httpWwwnaturecomf0cusrnaianimationsin deXhtml quotWitquot B i a Gene therapy is a technique for correcting defective genes responsible for disease development Gene knockin Gene knockout Selective reverse mutation Control regulation turn on or turn off RNAi 0 Gene function study Gene knockout eX Knockout mouse RNAi can be used to turn off target genes selectively Why a computational approach 0 This project is one of good examples of crossdisciplined problems 0 In vitro experiments cannot produce results that can be analyzed in details 0 In silica approach could generate a lot of false positives Combining two approach we may obtain a better working model Developing a siRNA designing tool 0 Provide comprehensive effects of a selected siRNA especially potential side effects OffTarget Effects 0 Need to eliminate the number of false positives as many as possible Problems 0 Overview 0 In vitro assay does not have good resolutions for what really happened in a cell gt either cells died or no apparent changes after target gene silencing Even the cost of in vitro assay make the problem worse OffTarget Effects 0 Arti cially synthesized siRNAs may shut down other genes as well Badly designed siRNAs can turn off more than 9000 genes 0 Computational analysis for OTE has been focused on sequence matching procedure However OTE analysis requires extensive sequence comparisons over entire chromosomes gt computationally expensive Problems In Vitro Approach 0 siRNA Designing problem A popular method is to select a portion of a target gene blindly and test it To select the best region of a target gene multiple regions might be tested The cost of siRNA testing is very expensive 0 More than few thousand dollars to test 96 siRNAs It is not feasible to gather OTE information from in vitro test It is hard to design a siRNA Without OTE information 0 Simple comparisons Computational complexity is not a problem but the size of data and output matters Computational time The cost is directly related to the length of a target gene and number of genes to compare Pr on 0 Computational time for OTE Analysis For the most simplest case only consider perfect match and 21mer A target gene select length of n characters Number of potential siRNAs from average length of 500 characters n Number of 21mer from a siRNA 1000 Total number of 21mer from the target gene 1000n A pool of nmer generated from target genes Other genes in Chromosome General Setting Length of s Number of fragments 2 x s Pr 03901 9 Running cost for straightforward comparisons Assume that the total number of genes in a species is g and average length of these gene is Z Basically each 21mer must be compared Z times on each gene lOOOnZ comparisons are repeated g times So total running time is lOOOnZg for one gene For a genome Wide analysis it will take 1000nlg2 gt 0g2 0g4 0 Computational representation of siRNAtarget binding Each gene g x in the input space I consisting of a sequence of characters drawn from the alphabet A a9 c g t9 A l 4 N4 nmer feature space C133 9a gxaeAn Kgxgy lt Zxgx ixgygt 2 T Cl 0 Example 01 aacgac and 02 aacgtgg using 3mer n3 exact match Dix 01 61610 acg cga gac 133102 6161076108 cghgtg tgg 0DF4 Where F is the number of nmers in the genome 40X10660X106 and D is amount of nmers to be compared close to F This is basically similar to 0n3 0 Their improvement Ignore nmers having zero occurrence gt a gene can be represented in the feature space compactly Sort features with nmers and used a balanced binary search tree 0kxl0gF Where kx is the number of nmers in a gene amp Constructa Q balanced binary search tree 9 o b d O l l l Create a list of all possible n mer d o What problems are still remaining 0DF4 vs 0kxlogF kx could be huge in genome Wide comparisons close to F or D Generate a storage problem to hold a balanced binary tree into memory number of nodes will be 4n in the worst case 399 Parallelize comparison step 0 Improve on D nmers to match or F nmers in genome Create a dictionary for nmers to be scanned D through F Number of node 4n1 Each node has 4 pointers for next character Null pointer indicates either string ending or no string matching on the pointer path 0F However this works for only perfect match possibly one mismatch Once arbitrary number of mismatches insertions and deletions is allowed the tree will be nearly full Possible Improvements Better handling for mismatch insertion and deletion cases 0 Some of perfect matches do not have any effect on target gene silencing Why Target mRNA might be in thermodynamically stable state gt formed secondary structure to obtain minimum energy level Instead of using simple sequence similarities consider hybridization energy OligonucleotidemRNA hybridization free energy prediction Stealing ideas from Biochemistry Gibbs Free Energy 0 In thermodynamics the Gibbs free energy is a thermodynamic potential which measures the useful work obtamable from a closed thermodynamic system at a constant temperature and pressure 0 In translation macromolecules like mRNA has tendency to release energy to maintain thermodynam1cally stable state A general rule of thumb 0 Every system seeks to achieve a minimum of free energy 0 A quantitative measure as to how near or far a potential reaction is from this minimum 0 The change in Gibbs free energy AG is negative then it refers that such reaction is favored and Will release energy If AG is positive then the reaction requires energy to be happened IV E a Free Energy Prediction I The overall free energy of siRNA binding Free energy of RNA unfolding O A AGnf01d no AGquot total restruct The energy to refold the Free energy for the intermolecular interaction mRNA into a new minimum energy structure quota Decision of a proper free energy threshold Experimental data this is done OTE results from a computational analysis By combining above results the most realistic hybridization energy level could be obtained O 0 T 2 r Hr w 4 life m 3 Qiu et al A computational study of offtarget effects of RNA interference Nucleic Acids Research 2005 Vol 33 No 6 18341847 Walton et al Thermodynamics and Kinetic Characterization of Antisense Oligodeoxynucleotide Binding to a Structured mRNA Biophysical Journal 2002 Vol 82 366377 Hi is 392 V i 5 E 39 3914 3 g 4 Sec 011 Dman Routing Techniques for Ad Hoc Networks X13 Kevin Su Mobile Ad hoc NEMrks WTS Secutity Association Graphs SASS Analyze existing secure routing protocols 183113 BAGS Two new attacks A new Secure Ora demand Routing SOR mechanism Simulations Conclusion Cmunicatim via wirclcss We unreliable No centralized controller and in astructure Nodes can performs the roles of both hosts and routers Limited processing capability and energy budget Dytmxnic network topologg y39 Intrinsic mutual trust Proactive table dxzivetx approaches DSDV Destination Sequenced Distance Vector 018R Optimized link Static Routing Reactive on derrxand approaches DSR Dynamic Source Routing AODV Adhoe On dennand Distance Vector Hybnid approaches ZRP Zone Routing Protocol I g9 3350 lug 39 90 3E 50 as 99 93quot 9 33 3E9 Riga EARN Egg Bl gr Mfg Ag PUP J r L guanoquotmag 0H L r L H 90 gay 0H9 man go MR pg I nalitywag lgmg EDUHPVOL uh uh L WM Em inog aHKVMVI ngv ggwlm gaguogg g Hal 15300 Mvui m 0 Raniquot gnu nahEV g ml lg m3 Myron UAVF mu 0 053730 URN Hon 06 Egg 9 33V I waxHBO ggg nuu I HM g ggg 06 a guyquot no gain i A gnuu no RHquot ngn Prov mu nu nw EVERup W gab I 3660 aquot 855 985 EU 8 Rug 0 3quot Ho EWGHHHU E0 gran sign Route Discovery Broadcast transmission gt Represents transmission of REQ 901 gtU lo 39 85Q II I 20139 u 30720 ung as a 33 I Pan 0 30 10355 an u VIKU I I 01 AM uquot 39nq In again an Ava0 a 5 to 035 I39VEIna 93 n 10 Iialanu 6050 2350 00 8 302 a an Oil39l bnh II I RC3 runsht REP RC3 runsht REP RC3 runsht REP A Ropmscntn a link an thi forwurd path g g an3 E 9 g kwng Innahrquot g 50 gawk Enn nu 9930 g Banning haw canning paggv g 38 quotgna mm SUV Egg i nu gv moanquot 30 Om 00 39 g un u I wrathquot gunman nan8 HOEgm Hogan Educ nag v fuwu A H u v39nuun un raga HBOHA 2539 080 I 5013500 lamhagu h nonhumanBa hogan 3905500 0K 50an 39 39 ma a053 Eng Hoganquot Ii Hum g Margawa Mobile Ad hoc NE Norks WTE Security Association Graphs SAGs Analyze ends ng secure routing protocols using SAGS Two new attacks A new Secure Can demand Routing SOR mechanisrn Simulations Conclusion Gnu way Hash Function MDS 128m39t RIPEBdDl f SHA l etc Syrxxmetxical Cryptography shared key AES RC4 etc Public Key Cryptography public private keys RSA etc Block cipher Message authetl cation code MAC digital sigxlamm I Wt associaiians in a mntixag rotatel can be represented by a graph G V V V n denotes the set of network nodes I E m denotes the 3610deth lirxka I Four basic types of directed links I Point tmpoint I Hop by hop I Cumulative I Mnltipoint I Directed links going oorn left to 213 ht show secuni mechanisms used in the route request stage ty I Dimctexi links going om tight to left Show security mechaxnisxns used in tile route reply stage l Li111 11361 ill seC LIl itIy39 association grapl ls l c11111111atiVv39e li111 4 I 11613 1337 110 1i1 3 1 oiiit ttb 13 Oillt 1i111lt 111111ti130 i111 1 i11k digital sigrlatLlre 118811 fLii lCtiOii MAC block cipher CLliliLllative MAC patli ertcrypted with block cipller Fig 2 LirLk Labels L ILSed SAGS cont Each hash MAC encrypted information or digital sigrlature is considered as a security code A point tO point link is a normal directed link between two nodes The node at the start of the link generates a security code and the end node veri es it lteg block cipher A hop by hop link can be viewed as two or more point to point links connected in a sequence it passes through one or more intermediate nodes between the end points A hop by hop litik denotes that each node in the path lteg one way hash chain receives the security code from the previous node except the rst node Which creates the initial security code veri es it replaces it With a new security code based on the received one d passes the new code to the next node in sequence except the end node Which does not pass it to any other node SAGS cont A cumulative link is similar to a hop by hop link except that the edges at all but the last hop are undirected A cumulative link denotes that each node in the path eg cumulative MAC receives the security code from the previous node except the first node Which creates the initial security code replaces it With a new security code based on the received one and its information passes the new code to the next node in sequence except the end node Which verifies it A multipoint link denotes the situation Where a node creates an authentication code and two or more nodes verify it eg digital signature Mobile Ad hoc NE Norks WTE Secutity Association Graphs SAGS Analyze endsting secure routing protocols using SAGs Two new attacks A new Secure Can demand Routing SOR mechanisrn Simulations Conclusion Based on ARAN 2002 SAODV Perkins 2002 I Based on DSR SR Padmmbhan 2002 Ariadne llIn 2005 endaixA Acs 2006 III S signs the BBQ packet it initiates Each aode verifies the signamre of the previous node replaces the signature of tile previous if it is not time source Witt its sigrtamre of tine packet and retransmit it D veri es the signatures of its previous hop and the source Route reply stage use me same secure mechanisms as in route request stage S A B C D C B A ARAN cont ROUTE REQUEST D certs N tlt REQUEST D certs REQUEST D certs REQUEST D certs RQUTE ROUTE ROUTE ROUTE ROUTE RQUTE ROUTE RE PLY RE PLY RE PLY RE PLY certD N certD N certD N certD N N tK K3 certA N certs N tllt lltEr certc tlt5 tlt5 Kg certc t K5 Kg certB 13 K5 K2 certA quotI neat 33W Digital signatures to authenticate the non mumble elds source id lest id some seq of REQS One way hash chains to secure the turntable elds of REQs Route reply stage use time same secure mechanisms as in route request stage SAODV cont Seqs r D Oldsqu rh C N Kg 0 hm RREQ id 5 seqs D oldsqu ho I39Nquot Kg 1 h gjr RREQ id 5 squ D oldsqu ho N Kg 2 FIN2 RREQ id 5 seqS D oldsqu ho N Kg 3 hIN3 RREP D seqo 5 lifetirne 75 N K5 0 F13 RREP D squ 5 lifetime ha N K5 1 hg7gt lifetirne fr N K5 2hampL2 seqo 5 lifetime hf N K5 3 hl Lg RREQ id 5 RREP D Squ 5 I R REP D SRP Requires security veri cation only between S and D using MAC fog REP pa1ltets c SRP 133quot P alJa lilllitr atos 311d I Iaas Ariadne TE SLA ho MACKSDazREQUEsr 5 D idti REQUEST S D id ti ho J 1 PM HA he MA MACKAEREQUEST 5 D idtr hr A 0 REQUEST 5 D id ti h1 A MA 12 HB H7 ME MACKEVI REQUEST 5 D idfi 112 A 3 MA u REQUEST 5 D id ti I12 A B MA MB 13 HC 12 MC MACKQI REQUEST 5 D idtr39 h3 A 8C MA MB REQUEST 5 D id ti h A 3C MA MB MC MD MACKDS REPLY D 5 ti A 3C MA MB MC REPLY D 5 ti A 8 C MA 1548 MC MD REPLY D 50 A B C MA MB MC MD KCH REPLY D 5 If A 3 C MA M3 MC MD KCH KB REPLY D 5 ti A 8 C MA A43 MC MD KC KB KAN d AAIiaclnLeiTE SLAA Ariadne MAC Shared pair Wise keys are used for all nodes D instead of S veri es the MAC value created by intermediate nodes e Al i a Ille IV39IJA CT Ariadne 10W overhe ad Cl s I 397 r f H E d mi 7 9 vquot 7 uw quot QR J Q39 f Aliadlle T ESLA LOVV overllead g Al iadlleiMZ XC Lo w overllead endairA C7 SisD C SigD C 9i5339D 4 11 e11 da irgk Mobile Ad hoc NEMrks WTS Secutity Association Graphs SASS Analyze existing secure routing protocols 183113 BAGS Two new attacks A new Secure Ora demand Routing SOR mechanism Simulations Conclusion TWO New Attacks Attack 1 Reactive Attack If REQS carry the path traversed in clear text E1 Upon receiving a REQ Y can check if the path already contaitls X and query X the authentication code X i generated E Y tunnel the REP back to X Works in SRP Ariadne and endairA TWO New Attacks cont Attack 2 Proactive Attack Upon receiving a REQ a malicious node lteg X if not close to D tunnel actual info in the REQ to other malicious nodes eg Y A malicious if it is neighbor of D discard legitimate REQs and wait for the first tunneled info from another malicious node Y tunnels the corresponding REP back to X Requires maintain paths to each other Works in SRP Ariadne endairA ARAN and SAODV Mobile Ad hoc NE Norks WTE Secutity Association Graphs SAGS Analyze ends ng secure routing protocols using SAGS Two new attacks A new Secure On dernand Routing 80R mechanism Simulations Conclusion Incorporate several designs feamres Autismexmticate intennediate nodes Hide paths taken by route request packets Propagate request lstex Avoid route selection based on hop count desigxn alternatives Route may be disserxninated to all nodes in time route only to source only no des nacion Either source or destination veri es the securing code generated by the intetrnediam nodes SOR DV destination Veri es REQ X RI 2Q 539 S JSEPX vallere 118 AriII CSD S D a S3 a EPS Am9 311d X 4 539 f REE S39 D Pill 5391 AilID Krllere EkufzilCIfSD REP S39s fE D PLgtIS39T REQ X erlel e S39 311d C 39 7 39 vvitll Chad39s RE RY REP S D AID EPX imrllele D 4 C D a11c1 E RX S39 Da r39a C IBHD 3quot E KXSFHX EPFHX xvitll E PD Mobile Ad hoc NEMrks WTS Secutity Association Graphs SASS Analyze existing secure routing protocols 183113 BAGS Two new attacks A new Secure Ora demand Routing SOR mechanism Simulations Conclusion Simulation Setup Nllxnb er Node S pe e 1 Node IvIobiler P8115 e Tilne Fie 1d S ize Raidio R2111 ge IxI AC Nulllb e 0 f Tra ic Pa irs Tra Eic L o a cl Data Pa cket Pay lo a d Lth B Reply Chara ti 011 St IIlitia 1 RE Q Tilneo th Iviax i11111111 RE Q TilIleouI R0111 e Ca 0116 S ize Hasll lengtll Cipher Block size 50 1 19111quots I VIO di e d Randonl XRVajy39p o i11t 0 90 0 seconds 1500 111 X 300 In I 1300 111 X 800 In I 100 300 Kbps CBRKUDP 500 by tes 2 LIbps 1 O 111d Illisecontds 05 seconuds 1 O secontds 32 renames V39u39itll FIFO replacemlmellt O O l 739 Simulation Re 8 ults DSR No Ariadne Hi0 Aria d ne LoOvd 2 U 31 gt 1 I 2 I Q 4 CI x 0 CU CI 200 300 400 500 600 700 800 900 Pause Time s Simulation Results cont DSR NoOpt Ariadne HiOvd Ariadne LoOvd SOR DV SOR SV 400 500 Pause Time s 30C Simulation Results cont Attack quotC CD U E c D E CG D quotc 2 G 1 C 3 8 N umber of Attackers Simulation Results cont Ariadne SV Ariadne Ariadne V a 33 my CD 0 30 031 x 0 mg 02 D o 0 GJ 50 Li 8 M umber of Attackers Simulation Results cont Ariadne Attack SV Attack Ariadne Attack Ariadne Attack V Attack on 4 13 D a 2 9 2 m E a CD gt O H O 8 N umber of Attackers Semtity Associa on Graphs SAGS facilitate analyses of e s ng secure rou ng protocols NO new attacks were presernted which can be launched on existing seculre routing protocols A new secure mechanism 8011 was proposed and simulations showed the effecdvetless of the proposed approach References Ilt Sanzgiri B Dahill B Levine and E BeldingiRoyer A Secure Routing Protocol for Ad hoc Networks In Prooeedingy ofIEEE ICNP 2002 C E Perkins E M BeldingiRoyer and S R Das Secure Ad hoc One demand Distance Vector SAOD V Routing A CM SICMOBILE Mokie Coiizlbzle g and Communiocztionx Review 1067107 2002 P Papadmitratos and Z Haas Secure Routing for Mobile Ad hoc Networks In Prooeedingy oftbe 5C5 Communication Nefwor x and Dixtribitz ed yJ Z einJ ModeZing and Simulation Conference CNDS 2002 Jan 2002 YihiCIhun Hu Adrian Perrig and David B Johnson Ariadne A Secure Onideman Routing Protocols for Ad Hoc Networks Wire 55 Networ x 1 1 17221738 2005 G Acs L Buttyan and I Vajda Provably Secure Onidemand Source Routing in Mobile Ad hoc Networks IEEE Tranxczoz ionx on Mokie Coiizlbzle g 511153371546 Nov 2006 Question 5 bit strl length nblt be strings I H is a lousy compression anction Co inionl hxh x fox some input a Result of bathing should look 39 39 39 I Two Pro satirgnvnma W madam ye 01 hard to nal say a such that thy Coniaioa Easiesthch hand to xmd x x such that hxhxquot 0W1 b1 baa bbaquot s bu 6025 80111 arbitrary value 1 1 Hb1 for all 1lt i ltn Given bi it is easy to veri r the authenticity of bi ifj lti W aumuuuon cod massage MACCKEYmemue 39 Bob Recomqu MAC and varl eo whcthar It Is equal to the MAC attached to the mesaa Alice Integrity and authentication only someone who knows KEY can compute MAC for a given message


Buy Material

Are you sure you want to buy this material for

25 Karma

Buy Material

BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.


You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

Why people love StudySoup

Jim McGreen Ohio University

"Knowing I can count on the Elite Notetaker in my class allows me to focus on what the professor is saying instead of just scribbling notes the whole time and falling behind."

Janice Dongeun University of Washington

"I used the money I made selling my notes & study guides to pay for spring break in Olympia, Washington...which was Sweet!"

Steve Martinelli UC Los Angeles

"There's no way I would have passed my Organic Chemistry class this semester without the notes and study guides I got from StudySoup."


"Their 'Elite Notetakers' are making over $1,200/month in sales by creating high quality content that helps their classmates in a time of need."

Become an Elite Notetaker and start selling your notes online!

Refund Policy


All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email


StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here:

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.