25 views

Uploaded by Kalolo Sadath

Wireless networks includes many networks from WLAN to GSM, GPRS, VSAT, UMTS, LTE and DVB

Wireless networks includes many networks from WLAN to GSM, GPRS, VSAT, UMTS, LTE and DVB

Attribution Non-Commercial (BY-NC)

- Crack PMP in 45 Days
- Wireless Networking
- Iub Over Satellite.ppt
- Tech Specifications Vsat
- PMP Processes
- Speech Recognition Using Hidden Markov Model Algorithm
- Huawei 3G Iub Satellite Transmission Configuration in RNC
- Satlink 2000 Vsat - Rev v(1)
- Ferrari e Galves - Construction of Stochastic Processes, Coupling and Regeneration
- ECE697AA 08-10-31 Queuing Systems I
- Voice Over VSAT
- 160
- s to Chas Tic Composer
- Physica A. Szilard
- Visuomotor characterization of eye movements in a drawing task
- TCP_Analyze_Sequence_Numbers - The Wireshark Wiki
- DYNAMIC POWER MANAGEMENT FOR NON-STATIONARY SERVICE REQUEST
- wsphereg
- Exam Problems
- TEMS Investigation KPI Definitions -- HTTP

You are on page 1of 51

Rasmus L. Olsen

Overview of course

1. 2. 3. 4. 5. Wireless Systems: GSM,GPRS and 3G Reliability aspects in Service and Context management architectures WiMax and Radio over Fiber Networks - 4G QoS Performance aspects and performance modelling

Content Motivation & Background Performance Analysis in Wireless Settings Review of Basic Concepts: Random Variables, Exponential Distributions, Stochastic Processes Birth-Death Processes, Discrete and Continous time Markov M/M/k/k type queueing models (revision) Kendall Notation, M/M/k/k type of queue models, steady-state solution Erlang-B formula Wireless Link Models IID errors, correlated errors, Gilbert Elliot Model Application: Queuing model with varying channel Multiple queues, multi-class models Simple Traffic Models Traffic measurements and characteristics ON/OFF Markov models Summary/Exercises

chains

Advantages of Packet-Based Transport (as opposed to circuit switched)

Flexibility Optimal Use of Link Capacities, Multiplex-Gain for bursty traffic Buffering/Queueing at routers can be necessary Delay / Jitter / Packet Loss can occur Overhead from Headers (20 Byte IPv4, 20 Byte TCP)

queueing

Drawbacks

Main motivation for Performance Modeling:

User Plane QoS/Network Performance

End-2-End Packet Delay (in particular interactive applications) Delay Jitter Packet Loss Throughput/Goodput e.g. Video/Voice Quality (depending on codecs) Call Setup Delays Fraction of blocked Calls Failure probablities of entities Downtime distribution Dropped Calls Delayed / Lost packets

Reliability Aspects

Behavior at Handover

Ultimate goal of service provisioning: user satisfaction Focus in this course: Network and application aspects, i.e. L2-5 and application layer Relevant functionalities:

PHY Layer Bit/Symbol transmission Throughput Symbol error probabilities (channel

User

User perceived QoS

User Environment

Middleware

Application QoS

Network QoS

conditions, interference)

L4: Transport: TCP, UDP, RTP/UDP L3: Network Layer: IP L2: MAC/LLC L1: PHYS

Link Layer (L2)

Medium Access Delays Collisions/unsuccessful transmissions Fragmentation Forward error correction (FEC) and error detection (CRC) Link-layer Retransmission Mechanisms (ARQ) L2 scheduling, switching, buffering

Path selection (routing) Processing delays (e.g. for routing table lookup) L3 buffering, scheduling, buffer management (RED) [L3 fragmentation]

Transport Layer (L4)

Multiplexing/de-multiplexing (UDP/TCP) Error detection/checksums In-order delivery, sequence numbers (TCP) Acknowledgements and Retransmissions (TCP) Flow/Congestion Control (TCP)

Application Layer/Codecs

FEC/CRC Application Layer Retransmissions Application Layer sequence numbers

All Layers

Increased volume due to headers

Challenges in IP networks:

Multiplexing of packets at nodes (L3) Burstiness of IP traffic (L3-7) Impact of Dynamic Routing (L3) Performance impact of transport layer, in particular TCP (L4) Wide range of applications different traffic & QoS requirements (L5-7) Feedback: performance traffic model, e.g. for TCP traffic, adaptive applications

HTTP L5-7 L4 L3 L2

Challenges in Wireless Networks: Wireless link models (channel models) MAC & LLC modeling RRM procedures Mobility models Cross layer optimization Analysis frequently with stochastic models

TCP IP

LinkLayer

Random experiment with set of possible results Axiomatic definition on event set P()

Definition: X: R; Pr(X=x) = p(x)

Probability density function f(x), Cumulative distribution function F(x) = Pr(Xx), Reliability function (complementary distr. Function) R(x)=1-F(x) = Pr(X>x)

Number of packets that arrive at the access router in the next hour (discrete) Buffer occupancy (#packets) in switch x at time y (discrete) Number of downloads (mouse clicks) in the next web session (discrete) Time until arrival of the next IP packet at a base station (continuous)

Important example: Geometrically distributed RV X (parameter p, 0<p<1)

Pr(X=n) = (1-p)n-1p, n=1,2,3,... Expected value, E(X) = 1/p, and variance, Var(X) = (1-p)/p2 Example:

1. 2. 3. 4. 5.

Initialize counter c=0 c=c+1 throw a biased coin (heads with probability p) If coin shows heads go to step 2 Otherwise stop and output c

Single parameter: rate Density function f(x)= exp(- x), x>0 Cdf: F(x)=1-exp(- x), Reliability function: R(x)=exp(- x) Moments: E{X}=1/ ; Var{X}=1/ 2, C2 = Var{X} / [E{X}]2 = 1

Memory-less: Pr(X>x+y | X>x) = exp(- y) Properties of two independent exponential RV: X with rate , Y with rate

Important properties:

Inter-arrival time process: Xi Counting Process: n-1 N(t) = max{n | i=1 Xi t} , alternatively Ni() = N(i) N([i-1])

t (t)

Important Example: Poisson Process

Assume i.i.d. exponential packet inter-arrival times (rate ): Xi:=Ti-Ti-1 Counting Process: Number of packets Nt until time t

Properties:

Merging: arrivals from two independent Poisson processes with rate 1 and 2 Poisson process with rate (1+ 2) Thinning: arrivals from a Poisson process of rate are discarded independently with probability p Poisson process with rate (1-p) Central Limit Theorem: superposition of n independent processes results in the limit n in a Poisson process (under certain conditions on the processes)

Definition

State-Space:

finite or countable infinite, E={1,2,...,N} (N= also allowed)

Transition probabilities: pjk=Pr(transition from state j to state k) Xi = RV indicating the state of the Markov process in step i Markov Property: State in step i only depends on state in step i-1 Pr(Xi=s | Xi-1=si-1,Xi-2=si-2 ,..., X0=s0 ) = Pr(Xi=s | Xi-1=si-1)

Initial state probabilities (Step 0): 0 Probability of state-sequence s0 ,s1 ,...,si: Pr(X0=s0 ,X1=s1 ,..., Xi=si ) = ... Pr(Xi=k)=j [Pr(Xi-1=j)*pjk] i = i-1 P

Properties

homogenuity: P independent of step i Irreducibility: each state is reachable from any other state

(in potentially multiple steps)

Steady-state probabilities

= limi i Limit exists and is independent of 0 if Markov chain irreducible and aperiodic Aperiodic & positive recurrent = ergodic is probability distribution

Defined by

State-Space: finite or countable infinite, E={0,1,2,...,K} (K= also allowed) Transition rates: jk

Holding time in state j: exponential with rate kj jk Transition probability from state j to k: jk / lj jl =: jk / j

Xt = RV indicating the current state at time t; i(t):=Pr(Xt=i) Markov Property: transitions do not depend on history but only on current

state t0<t1<...<tn , i0,i1,...in ,j E

0 1 2

Steady-State Probabilities:

i := Pr(Q=i) = 0 k=0i-1 k / k=1i k 0:= (1+i=1(k=0i-1 k / k=1i k))1

Exercise 3: multiple

server queue, see last slide (15min)

Content

Motivation & Background

M/M/k/k type queueing models (revision) Kendall Notation, M/M/k/k type of queue models, steady-state solution Erlang-B formula Wireless Link Models IID errors, correlated errors, Gilbert Elliot Model Application: Queuing model with varying channel Multiple queues, multi-class models Simple Traffic Models Traffic measurements and characteristics ON/OFF Markov models Summary/Exercises

Performance Analysis in Wireless Settings Review of Basic Concepts: Random Variables, Exponential Distributions, Stochastic Processes Birth-Death Processes, Discrete and Continous time Markov chains

K channels Users allocate one channel per call for certain call duration If all channels are allocated additional starting calls are blocked How many channels are necessary to achieve a call certain maximal blocking probability?

Common Model Assumptions: Calls are arriving according to a Poission Process (justified for large user population, limit theorems for stochastic processes) with rate Call durations are exponentially distributed with mean T (okay for voice calls)

X/Y/C[/B] Queues (example: M/M/1, GI/M/2/10, M/M/10/10, ...) X: Specifies Arrival Process Y: Specifies Service Process (M,G(I),...) C: Number of Servers B: size of finite waiting room (buffer)

[also counting the packet in service]

If not specified: B=

FIFO: First-In-First-Out (default) Processor Sharing: PS Last-in-first-out LIFO (preemptive or non-preemptive) Earliest Deadline First (EDF), etc.

M/M/1 queue

Poisson arrival of packets (first M Markovian) with rate Exponentially distributed service times of rate (second M) Single Server (1) FIFO service discipline Qt = Number of packets in system is continuous-time Markov Process

Infinite buffer

Derived Parameter:

Utilization, = / : if 1, instable case (no steady-state q.l.d)

Performance Parameters

Queue-length distribution: (t) , steady-state limit: =lim (t) (if <1) Queue-length that an arriving customer sees Waiting/System time distribution

Birth-Death Process

Probability of idle server: 0 = (1-) Probability of i packets i := Pr(Q=i) = (1-)* i, in queue: where = / <1 Average Queue-length: E{Q} = /(1-) Average Delay (System Time): E{S} = E{Q}/ = 1/(-)

Poisson arrival of packets with rate Exponentially distributed service times of rate Single Server Finite waiting room (buffer) for K packets

Suitable e.g. for modeling bottleneck link in packetbased wireless networks [Full network models: see Markov Models II lecture]

Additional performance metric

buffer occupancy B or higher)

Probability of i packets in queue Probability of packet loss:

i := Pr(Q=i) = (1-)/(1- K+1) * i, where = / 1, i=0,,K

Probability of i calls active i := Pr(n=i) = 0 (T)i /i! , i=1,,K where 0 = 1/[(T)i /i!] (sum taken over i=0 to K) Probability of blocked call: [also known as Erlang-B formula]

p(Blocking) = K = 0 (T)K /K!

Finite buffer (size K)

Content Motivation & Background Performance Analysis in Wireless Settings Review of Basic Concepts: Random Variables, Exponential Distributions, Stochastic Processes Birth-Death Processes, Discrete and Continous time Markov M/M/k/k type queueing models (revision) Kendall Notation, M/M/k/k type of queue models, steady-state solution Erlang-B formula Wireless Link Models IID errors, correlated errors, Gilbert Elliot Model Application: Queuing model with varying channel Multiple queues, multi-class models Simple Traffic Models Traffic measurements and characteristics ON/OFF Markov models Summary/Exercises

chains

Independent errors

Simplest error model: every bit (L2 frame, packet) independently experiences the same error probability pb (pf , pp) Number of erroneous bits: Bernoulli distributed Probability of correct frame transmission for frame-size s Assuming simple ARQ (repeat until successful)

Without FEC With FEC (assuming that n bit errors can be corrected) Number of needed attempts geometrically distributed with parameter pf Corresponds to geometrically distributed frame-delay in slotted systems

transmission error)

Steady-state error probability Error burst length: geometric

Correlated Errors

For correlated channel errors: Markov modulated error model:

While current state i: packets corrupted with probability i After each packet: state transition with probability pik

Easiest Case:

2-state model: Gilbert-Elliot Long-term packet error rate

Main property

In each state s E, an observation symbol from some alphabet V is generated probabilistically The underlying state cannot be observed, only the sequence O=[O1,O2,...,OT] of generated symbols

E: state-space (discrete, finite/infinite), E={1,2,...,N} V: set of possible observation symbols (discrete for now), V={1,2,...,M} 1: initial state probabilities at step 1 P: NxN matrix of state transition probabilities pij = Pr(Xk+1=j | Xk=i) B: NxM matrix of symbol generation probabilities: bij = Pr (Ok=j | Xk=i)

namely each column of B contains at most one non-zero element

The observation sequence above can be produced by the following state sequences. 532532 531212 432532 431212 312532

But which, and what is the probability of observing one of the sequences?

Given the parameters of the model, compute the probability of a particular output sequence, and the probabilities of the hidden state values given that output sequence. This problem is solved by the forwardbackward algorithm. Given the parameters of the model, find the most likely sequence of hidden states that could have generated a given output sequence. This problem is solved by the Viterbi algorithm. Given an output sequence or a set of such sequences, find the most likely set of state transition and output probabilities. In other words, discover the parameters of the HMM given a dataset of sequences. This problem is solved by the Baum-Welch algorithm or the Baldi-Chauvin algorithm.

Problem: Compute probability of observing a certain sequence o=[o1,...,oT] in a given HMM.

Generate all possible state-sequences of length T: q=[q1,...,qT] Sum up all Pr(o| q) weigthed by Pr(q) (total probabilities) Advantage is that it is a simple approach Disadvantage is the time complexity is O(TNT)

Given the observed sequence, moves forward through the sequence and calculates probabilities for next sequence element. Once reached through the given observed sequence, the algorithm traverse backwards similar, performing similar steps and does smoothing as well Advantage is the time complexity is good O(N2T) Disadvantage is the complexity

First (inefficient) approach (brute-force):

Generate all possible state-sequences of length T: q=[q1,...,qT] Sum up all Pr(o| q) weigthed by Pr(q) (total probabilities) Problem: Number of paths grows exponentially as NT More efficient (quadratic in N) approach: forward procedure Iterative method computing probabilities for pre-fixes of the observation sequence:

Similarly (and identifiers needed later): Backwards procedure t := [Pr(Ot+1=ot+1,...,OT=oT| Xt=1), ..., Pr(Ot+1=ot+1,...,OT=oT | Xt=N)] T =1 (vector with all elements = 1); t = (P * B(:, ot+1 )) .* t+1

t := [Pr(O1=o1,...,Ot=ot, Xt=1), ..., Pr(O1=o1,...,Ot=ot, Xt=N)] At step t=1: 1(i) = Pr(O1=o1, X1=1) = 1(i) b1,o1 [ Matlab Notation: 1 = 1 .* B(:, o1 ) ] tt+1 (t=1,2,...,T-1): t+1(i) = (jE t(j) pji) Pr(Ot+1=ot+1 | Xt+1=i ) t+1 = (t P) .* B(:, ot+1 ) Finally: Pr(O=o) = jE T(j) Computation can be illustrated in Trellis structure

Poor

E0={(n,P)}

E1={(n,G)}

Extension of M/M/1 queue model for channel-dependent service time, e.g. Gilbert-Elliot model with ARQ Two different cases

Good

Poor channel condition service rate 0 Good channel condition service rate 1 > 0 Channel changes only after successful transition Channel may change at any moment in time

Two dimensional Markov chain: E={n=0,1,2,} x {G, P}, Transition Rates? Solutions: Truncate and compute numerically or use Quasi-Birth-Death Process solution (Markov Models II, 9th semester)

Application to scheduling Scenario: GPRS type of system (TDMA), slotted (discrete time) Two users in radio cell: Each user has a downstream packet flow, assumed to be Bernoulli arrivals with parameter ai Downstream packets are stored in two different queues

User 1 with good radio condition, service probability p1 User 2 worse, service probability p2< p1

The scheduler needs to make a decision which queue to serve (non preemptive!)

a1 a2

p1/2

Discrete time Markov chain State-space: E={(n1,n2)} = G 0xP0 Extended state space (odd/even slot?) for RR Transition probabilities ?

Content Motivation & Background Performance Analysis in Wireless Settings Review of Basic Concepts: Random Variables, Exponential Distributions, Stochastic Processes Birth-Death Processes, Discrete and Continous time Markov M/M/k/k type queueing models (revision) Kendall Notation, M/M/k/k type of queue models, steady-state solution Erlang-B formula Wireless Link Models IID errors, correlated errors, Gilbert Elliot Model Application: Queuing model with varying channel Multiple queues, multi-class models Simple Traffic Models Traffic measurements and characteristics ON/OFF Markov models Summary/Exercises

chains

Frequently used: Several levels with increasing granularity

E.g. 3 levels: sessions, connections, packets Or: 5-level model:

Main objects contain zero or more embedded objects that the browser retrieves Correlated requests for embedded objects within retrieval of main object

HTTP Session (User A) HTTP Session (User B) HTTP Session (User C) Session Level

...

click

...

...

Dld. Phase K

Statistics:

Daily Profiles/Non-stationarity

Utilization of networks varies during the course of a day

x 10

8

Traffic in OD Flow 84

1.2 1.15 1.1 1.05 1 0.95 0.9 0.85 Mon Tue Wed Thu Fri Sat Sun

indication of stationarity periods of 4 hours resp. 1 hour wrt. Poisson process assumptions for the user session arrivals

Burstiness

On packet-level (corresponding to shorter time-scales)

Poisson assumption with constant rate for packet arrivals may be applicable only for highly aggregated traffic (core networks) In access networks: typically varying arrival rates, e.g.

High data rates in ftp download but less activity between downloads http: activities after mouse-clicks Video streaming: high data rates in frame transmissions Interactive Voice: talk and silent periods

Model types:

Inhomogeneous Poisson process: time-varying rate (t) ON/OFF models Hierarchical models

Inhomogeneous Poisson process

Piece-wise constant approximation Probabilistic discarding

Simulate Poisson process with rate m > (t) for all t For event at time t, discard with probability (m-(t))/ m

Parameters:

N sources, each average rate During ON periods: peak-rate p Mean duration of ON and OFF times

= p ON/(ON+OFF)

bursty traffic, when p >>

0 0 0

ON

1 1 1

OFF

Two dimensional Markov chain: E={n=0,1,2,} x {ON, OFF} Transition Rates ? Solutions: Truncate and compute numerically or use Quasi-BirthDeath Process solution (Markov Models II, 9th semester)

Here: case of two ON/OFF sources, mean packet delay For small utilization<0.1 (large nu): less than factor of 2 compared to M/M/1 queue For larger utilization close to 1: Approx factor of 10 for the given parameters [Ignore curves labeled with T>1 for now]

One possibility: Burstiness = Peak Rate of Source p / Average Rate of Source Transformed to range 0 (not bursty) to 1 (very bursty):

b:=1-/p

Investigating burstiness:

Vary p Keep #packets per ON period scale ON duration accordingly Extreme Cases:

b=0 (p= ): Poisson arrivals, no burstiness b=1 (p= ): Bulk arrivals

Bulk Arrival Models Queue-length at arrival instances increases not only by 1, but by a Random Variable B, the bulk-size Parameter set of model

Bulk arrival process, e.g. exponential with rate Bulk-Size distribution: pi (e.g. geometric) Service rate (single packet)

[Chaudhry & Templeton 83]:

Non-exponential models

Long-range dependent and self-similar properties in network traffic Indications for the need to include heavy-tail distributions (see e.g. Web However, Markovian models (with certain structure) can mimic such distributions

file sizes below, right graph)

chains

Exercises:

1. Exponential Distribution, see Slide 11 2. Continuous Time Markov Chains, Chapman Kolmogorov Equation, see Slide 16 3. Multi-server queue: One router in a network (to be modelled as M/M/1 queue) is apparently creating large delays. The admin decides to add a second routing processor to it. In principle, the admin now has the choice to upgrade to an M/M/2 system, but he could also configure separate queues for the new routing module (packets will be randomly directed to one of the two queues); alternatively, he could replace the routing processor with another one that is twice as fast. Which solution is performance-wise better? Plot the ratio of the mean queue-lengths for varying arrival rate lambda. 4. Correlated Errors: Given is a 2-state Markov modulated bit error model with parameters P=[0.4,0.6; 0.2, 0.8], 1=0.5, 2=0.

Compute the long-term average bit-error rate. Write a Matlab program to simulate the error sequence Ei. Compute the coefficient of correlation r(k) and a histogram of error burst lengths. Compare the latter with a geometric distribution. Furthermore, compare with an independent error model. Inhomogeneous Poisson Process: Write a MATLAB program that creates samples from a timevarying Poisson process with lambda(t)= 1+sin(2*pi*t/100). Estimate the mean and the variance of this sample series (inter-event times) using n=1e4 samples. Compute an estimate of the autocorrelation.

5.

- Crack PMP in 45 DaysUploaded bygolfmaniac
- Wireless NetworkingUploaded byPrajwal Rao
- Iub Over Satellite.pptUploaded bySiva Kumar
- Tech Specifications VsatUploaded byconsolate
- PMP ProcessesUploaded byRobincrusoe
- Speech Recognition Using Hidden Markov Model AlgorithmUploaded byInternational Journal for Scientific Research and Development - IJSRD
- Huawei 3G Iub Satellite Transmission Configuration in RNCUploaded byygyggitinezhad
- Satlink 2000 Vsat - Rev v(1)Uploaded bysudanone
- Ferrari e Galves - Construction of Stochastic Processes, Coupling and RegenerationUploaded byFelipe Contratres
- ECE697AA 08-10-31 Queuing Systems IUploaded byabarai_renji1978
- Voice Over VSATUploaded byPravesh Kumar Thakur
- 160Uploaded byFernando Salazar
- s to Chas Tic ComposerUploaded byJudap Floc
- Visuomotor characterization of eye movements in a drawing taskUploaded byGiuseppe Boccignone
- TCP_Analyze_Sequence_Numbers - The Wireshark WikiUploaded byKempa Red
- DYNAMIC POWER MANAGEMENT FOR NON-STATIONARY SERVICE REQUESTUploaded byapi-26871643
- Physica A. SzilardUploaded byIon Nicu
- wspheregUploaded bym4u4
- Exam ProblemsUploaded byVan Dungct
- TEMS Investigation KPI Definitions -- HTTPUploaded byAhmed
- The Crowding AlgorithmUploaded byMossad News
- Lab 3 Ip Spoofing and Denial of Service Ms Word15Uploaded bySaif Aamer
- A Machine Learning Approach to Building Domain-Specific Search EnginesUploaded byHarsh Prabhu
- Chord Recognition Using Duration Explicit Hidden Markov ModelsUploaded byjunshern
- RootUploaded bysaudade96
- Video Classification:A Literature SurveyUploaded byEditor IJRITCC
- JGIS_2012123123415192.pdfUploaded byAntonio Gonzalez Patricio
- Network RevisionUploaded byLongNguyễn
- Project Report DocumentUploaded byNiklasKokkola
- Pres Lanc2007Uploaded byIerulescu Adrian Alexandru

- 64 Interview QuestionsUploaded byshivakumar N
- FreeRADIUS Log Files Format Interpretation for WebCDRUploaded byKalolo Sadath
- TEMS Discovery Device 10.0 - Commercial PresentationUploaded byKalolo Sadath
- TEMS Discovery Device 10.0 - Commercial PresentationUploaded byphandangnam
- ch01securityUploaded byKalolo Sadath
- wireless communicationsUploaded byKalolo Sadath
- SAP Manuscript TemplateUploaded byKalolo Sadath
- Wireless channelsUploaded byKalolo Sadath
- telecom statisticsUploaded byKalolo Sadath
- queueUploaded byKalolo Sadath
- ICSE6105 Lecture1 Introduction&MySQLUploaded byKalolo Sadath
- ICSE 6245 Lab Session-Matlab ReviewUploaded byKalolo Sadath
- 1. Introduction to 3GUploaded byKalolo Sadath
- s3Uploaded byKalolo Sadath

- The ELEMENTSofprobabilityUploaded byHaydn Bassarath
- Forecasting 2011 - Time SeriesUploaded byMohit Gupta
- Terence C. Mills - The Econometric of Modelling of Financial Time SeriesUploaded byBruno Turetto Rodrigues
- ReliabilityUploaded byRianaNurFatimah
- AYUDANTÍA 15 NOV.docxUploaded byPamela Caniuqueo Flores
- e StudioUploaded byconijadeperro
- Lognormal DistributionUploaded byYog Kunwar
- Probability and StatisticsUploaded byPrakash Dhage
- maths-4Uploaded byRohit Patil
- Statistics and Probability - Midterm Examination (TOS)Uploaded byRaymond Richard Magan
- 103_2016_3_bUploaded bythomas
- MSc syllabus.pdfUploaded byRAJESH KUMAR
- pqt-anna-university-notes-rejinpaul.pdfUploaded byDavid Miller
- DISC 203-Probability and Statistics-Muhammad Asim (2).pdfUploaded byMinhal Malik
- ArchUploaded bya ayesha
- sta457h1Uploaded byAnonymous MaUfbbza
- SBA - SAS.pdfUploaded byAjay Kuruvilla
- TagliafichiUploaded byNgân Nguyễn
- Properties of EstimatorsUploaded bythrphys1940
- s3Uploaded byRasidh72
- Analysis of CovarianceUploaded byAnonymous HbCzNow
- MIT18_05S14_ps5.pdfUploaded byIslamSharaf
- Confidence IntervalUploaded byAjinkya Adhikari
- Insurance RiskUploaded byAtul Yadav
- HW02Uploaded byManuel Coxe
- Chapter 06Uploaded byricha
- first passage time for a barrier optionUploaded byRajarshi Sarkar
- 2006 04 26 Statistical Issues in Drug Dev. Peter DiggleUploaded byhubik38
- MA 2262 — PROBABILITY AND QUEUEING THEORY1.pdfUploaded byvelkarthi92
- 04_Data analysis and Probability theory.pdfUploaded bydeepak