Probability and Statistics with
Reliability, Queuing and Computer Science Applications: Chapter 6 on
Stochastic Processes
Kishor S. Trivedi Visiting Professor
Dept. of Computer Science and Engineering Indian Institute of Technology, Kanpur
What is a Stochastic Process?
Stochastic Process: is a family of random variables { X(t) | t ε T } (T is an index set; it may be discrete or continuous)
Values assumed by X ( t ) are called states.
State space (I): set of all possible states
Sometimes called a random process or a chance
process
Stochastic Process Characterization
At a fixed time t=t
1, we have a random variable X(t
1).
Similarly, we have X(t
2), .., X(t
k).
X(t
1) can be characterized by its distribution function,
We can also consider the joint distribution function,
Discrete and continuous cases:
States X(t) (i.e. time t) may be discrete/continuous
State space I may be discrete/continuous
Classification of Stochastic Processes
Four classes of stochastic processes:
discrete-state process chain
discrete-time process stochastic sequence {X
n| n є T}
(e.g., probing a system every 10 ms.)
Example: a Queuing System
m servers Queue (waiting station)
Random arrivals Inter arrival time distribution fn. FY
Service time distribution fn. FS
Interarrival times Y1, Y2, … (common dist. Fn. FY)
Service times: S1, S2, … (iid with a common cdf FS)
Notation for a queuing system: FY /FS/m
Some interarrival/service time distributions types are:
M: Memoryless (i.e., EXP)
D: Deterministic
Ek: k-stage Erlang etc.
Hk: k-stage Hyper exponential distribution
G: General distribution
Discrete/Continuous Stochastic Processes
Nk: Number of jobs waiting in the system at the time of kth job’s departure Stochastic process {Nk| k=1,2,…}:
Discrete time, discrete state
Nk
k
Discrete
Continuous Time, Discrete Space
X(t): Number of jobs in the system at time t. {X(t) | t є T} forms a continuous-time, discrete-state stochastic process, with
,
X(t)
Discrete
Discrete Time, Continuous Space
Wk: waiting time for the kth job. Then {Wk | k є T} forms a Discrete-time, Continuous-state stochastic process, where,
Wk
Continuous
Continuous Time, Continuous Space
Y(t): total service time for all jobs in the system at time t. Y(t) forms a continuous-time, continuous-state stochastic process, Where,
Y(t)
Further Classification
Similarly, we can define n
thorder distribution:
Formidable task to provide n
thorder distribution for all n.
(1st order distribution) (2nd order distribution)
Further Classification (contd.)
Can the nth order distribution be simplified?
Yes. Under some simplifying assumptions:
Independence
As example, we have the Renewal Process
Discrete time independent process {Xn | n=1,2,…} (X1, X2, .. are iid, non-negative rvs), e.g., repair/replacement after a failure.
Markov process introduces a limited form of dependence
Markov Process
Stochastic proc. {X(t) | t є T} is Markov if for any t0 < t1< … < tn<
t, the conditional distribution satisfies the Markov property:
Markov Process
We will only deal with discrete state Markov processes i.e., Markov chains
In some situations, a Markov chain may also exhibit time- homogeneity
Future of process (probabilistically) determined by its current state, independent of how it reached this
particular state; but in a non homogeneous case, current time can also determine the future.
For a homogeneous Markov chain current time is also not needed to determine the future.
Let Y : time spent in a given state in a hom. CTMC
Homogeneous CTMC-Sojourn time
Since Y, the sojourn time, has the memoryless prop.
This result says that for a homogeneous continuous time Markov chain, sojourn time in a state follows EXP( ) distribution (not true for non-hom CTMC)
Hom. DTMC sojourn time dist. Is geometric.
Semi-Markov process is one in which the sojourn time in a state is generally distributed.
Bernoulli Process
A sequence of iid Bernoulli rvs, { Y
i| i=1,2,3,.. }, Y
i=1 or 0
{ Y
i} forms a Bernoulli Process, an example of a renewal process.
Define another stochastic process , { S
n| n=1,2,3,.. }, where S
n= Y
1+ Y
2+…+ Y
n( i.e. S
n:sequence of partial sums)
S
n= S
n-1+ Y
n(recursive form)
P[ S
n= k | S
n-1= k ] = P[ Y
n= 0 ] = ( 1-p ) and,
P[ S
n= k | S
n-1= k-1 ] = P[ Y
n= 1 ] = p
{ S
n| n=1,2,3,.. }, forms a Binomial process, an example
of a homogeneous DTMC
Renewal Counting Process
Renewal counting process: # of renewals
(repairs, replacements, arrivals) by time t : a continuous time process:
If time interval between two renewals follows
EXP distribution, then Poisson Process
Note:
For a fixed t , N(t) is a random variable (in this case a discrete random variable known as the Poisson random variable)
The family { N(t) , t 0} is a stochastic process, in this case, the homogeneous Poisson process
{ N(t) , t 0} is a homogeneous CTMC as
well
Poisson Process
A continuous time, discrete state process.
N(t): no. of events occurring in time (0, t]. Events may be,
1. # of packets arriving at a router port
2. # of incoming telephone calls at a switch
3. # of jobs arriving at file/compute server
4. Number of component failures
Events occurs successively and that intervals between these successive events are iid rvs, each following EXP( )
1. λ: arrival rate (1/ λ: average time between arrivals) λ: failure rate (1/ λ: average time between failures)
Poisson Process (contd.)
N(t) forms a Poisson process provided:
1. N(0) = 0
2. Events within non-overlapping intervals are independent
3. In a very small interval h, only one event may occur (prob.
p(h))
1. Letting, pn(t) = P[N(t)=n],
For a Poisson process, interarrival times follow EXP( ) (memoryless) distribution.
E[N(t)] = Var[N(t)] = λt ; What about E[N(t)/t], as t infinity?
Merged Multiple Poisson Process Streams
Consider the system,
Proof: Using z-transform. Letting, α = λt,
+
Decomposing a Poisson Stream
Decompose a Poisson process using a prob. switch
N arrivals decomposed into {N1, N2, .., Nk}; N= N1+N2, ..,+Nk
Cond. pmf
Since,
The uncond. pmf
Non-Homogeneous Poisson Process (NHPP)
Poisson Process
Generalizing the Poisson Process
Non-Homogeneous Poisson Process (NHPP)
If the expected number of events per unit time, , changes with age (time), we have a non-homogeneous Poisson model. We
assume that:
1. If 0 t, the pmf of N(t) is given by:
where m(t) 0 is the expected number of events in the time period [0, t]
2. Counts of events in non-overlapping time periods are mutually independent.
m(t) : the mean value function. (x) :the time-dependent rate of occurrence of events or time-dependent failure rate
t(x) dx )
( t
m
N t k m t
kk e
m t P / !
k 0 , 1 , 2 , ...
NHPP(cont.)
Non-Homogeneous Poisson Process (NHPP)
Poisson Process
Renewal Counting Process
Generalizing Poisson Process
Poisson process EXP( ) distributed interarrival times.
What if the EXP( ) assumption is removed renewal proc.
Renewal proc. : {Xi | i=1,2,…} (Xi’s are iid non-EXP rvs)
Xi : time gap between the occurrence of (i-1) st and ith event
Sk = X1 + X2 + .. + Xk time to occurrence of the kth event.
N(t)- Renewal counting process is a discrete-state, continuous- time stochastic process. N(t) denotes no. of renewals in the
Renewal Counting Process
Renewal Counting Processes (contd.)
For N(t), what is P(N(t) = n)?
Sn t
More arrivals possible
tn
Renewal Counting Process Expectation
Let, m(t) = E[ N(t) ]. Then, m(t) = mean no.
of arrivals in time (0,t]. m(t) is called the
renewal function .
Renewal Density Function
Renewal density function:
For example, if the renewal interval X is EXP( λ ), then
d(t) = λ , t >= 0 and m(t) = λ t , t >= 0.
P[ N(t) =n] = e
–λ t(λ t)
n/n! i.e Poisson pmf
Alternating Renewal Process
Where:
Failure times T1, T2, … are mutually independent with a common distribution function W
Restoration times D1, D2, … are mutually independent with a common distribution function G
The sequences {Tn} and {Dn} are independent
1
0 I(t)
Operating Restoration
Time
Availability Analysis
Availability: is defined is the ability of a system to provide the desired service.
If no repair/replacement,Availability(t)=Reliability(t)
If repairs are possible, then above is pessimistic.
MTBF = E[D
i+T
i+1] = E[T
i+D
i]=E[X
i]=MTTF+MTTR
T1 D1 T2 D2 T3 D3 T4 D4 …….
MTBF
Availability Analysis (contd.)
Two mutually exclusive situations:
1.
System does not fail before time t A(t) = R(t)
2.
System fails, but the repair is completed before time t
Therefore, A(t) = sum of these two probabilities
t
Repair is completed with in this interval renewal
x
Availability Expression
dA(x) : Incremental availability
dA(x) = Prob(that after renewal, life time is > (t-x) &
that the renewal occurs in the interval (x,x+dx])
Repair is completed with in this interval
x
Renewed life time >= (t-x)
0 x+dx t
Availability Expression (contd.)
A(t) can also be expressed in the Laplace domain.
Since, R(t) = 1-W(t) or L
R(s) = 1/s – L
W(s) = 1/s – L
w(s)/s
What happens when t becomes very large?
However,
Availability, MTTF and MTTR
Steady state availability A is:
Taking the expression of sL
A(s) and taking the limit via L’Hospital rule and using the moment generating property of the LT, we get the
required result for the steady-state
A=MTTF/(MTTF+MTTR)
Availability Example
Assuming EXP( ) density fn for g(t) and w(t)
Non-Homogeneous Poisson Process (NHPP)
Poisson Process
Renewal Counting Process
Homogeneous Continuous Time Markov Chain
Homogeneous Discrete Time Markov Chain
Semi-Markov Process
Markov Regenerative Non-Homogeneous
Continuous Time Markov Chain
Compound Poisson Process
Generalizing Poisson Process
Bernoulli Process