Вы находитесь на странице: 1из 3

NAME:- AMIR KHAN

M.S.C ELECTRICAL

Assignment # 03

"Report on Patch Antennas Issues


MSc Electrical Communication Engineering

Submitted by: Amir khan Submitted To: Dr. Shahid Bashir Subject: Antenna Theory & Design Semester: fall 2011

The paper present methodology for selecting the traffic regulation parameters based on divide & conquer rule. It ha two parts selecting of envelop from source data and selecting parameters from it. This model is faster than the existing ones.

Introduction:-

NAME:- AMIR KHAN

M.S.C ELECTRICAL

The upcoming high speed network require high regulation scheme to serve data. Cruz bound traffic generation by source A(s) in Between [0, s]. The time varying function bound maximum traffic between [s-t, t] given mathematically as The traffic characterization depend on two mechanism 1) Admission control mechanism, which determines if sufficient network resources are available to support a new connection at a desired QoS. 2) Tram policing mechanism which monitors the traffic of a source in realtime and checks if the actual traffic submitted to the network conforms to the traffic characterization of the source. This model is a generalization of the model, where the traffic is described with a burst parameter and a rate parameter the model consists of m pairs, 1< i < m, and the traffic is limited by each pair, i.e., A(t) = min , If pairs m increases accuracy increase but system become complex we have to trade of between m and accuracy. The size, accuracy and speed are challenging ones. The benchmark for evaluating the accuracy of a traffic source is the empirical envelope of the source, which is the tightest traffic constraint function of the some and is defined by: E(t) is most accurate traffic constrain but its calculation in realtime is not practical. Although extrapolation reduce load of calculation result in rough E(t). This paper show new algorithm practically applicable & easy and better results.

OVERVIEW OF RELATED WORK:The previous work include Dual leaky bucket, concave hull, product etc it has two parts given as 1). Constructing E(t) from the traffic:E(t) is obtained by searching maximum burst size in traffic using following equation find total length but it is very much time consuming and complex computational.

2). Finding the model parameters from E(t):E(t) is approximated as linear piece wise model A(t). it is replaced by concave hull denoted by H(E).then H(E)is replaced by another piece wise linear function the idea behind it is to find difference between H(E) &traffic constrain function. For each pair iteration is done through search. This is also more economical & computational; also not guarantee finding optimum parameters.

OUR APPROACH TO CONSTRUCT THE EMPIRICAL ENVELOPE:We present method for exact value of E(t) for 1<t< for t> E(t) at equally spaced samples the repeating samples are excluded not to be computed. So E(t) construction is for t<. 1). Construction of E(t) for t < : Take instantaneous bit rate of source X(i)=1,2,3N. E(t) is find as for t=1,2,3.. < & N Sk vector of size 1* can be find for k=1,2,3..N. The E (i) can be computed as shifting elements of down & adding results X(k). Three steps are 1stly consider k=1 and initialize s & e as 2nd let K=k+1 by

NAME:- AMIR KHAN


Sk = (shift elements of Sk-1 down by one) 3rd repeat step till input for k=N The complexity is not like the others E. construction of samples of E(t) for t = The steps involved are 1) let k=1 initialize n*1 vector

M.S.C ELECTRICAL

:-

2) let k=k+1 update


3) update vector A

as

4) update 5) update
step 2 to 5 are repeated for k=2,3,4..N

OBTAINING THE

PARAMETERS FROM THE EMPIRICAL ENVELOPE

The approach is to divide &conquer, it is easy to divide large problems to smaller of the same types and add them after solution and efficient in time consumption Our sub-problem fitting a single to E(t) :We find & such that for 1<t<N & . The first condition confirm constructed model is concave and other show optimum value of parameters. In order to find closeness in A(t ) & E(t) error function can be given as We need to minimize only factor effecting parameters

Conclusion:We divided the problem into two sub problems and presented robust methods for solving each one of them In the first pan of this paper, we presented an algorithm for constructing the entire empirical envelope 01 samples of it for a traffic source. Our method extensively reduces the number of computational operations such that computation of the empirical envelope in practical applications becomes feasible. Our algorithm can be employed for traffic characterization of on-line traf6c sources in applications that use QoS renegotiation. We also showed that by selecting appropriate parameters one can construct an accurate model from using just samples of the envelope rather than the entire envelope. We showed that our sampling technique increases the speed performance without degrading the accuracy of the constructed model. In the second part of the paper, we presented a simple and robust method for finding the parameters of the model from the empirical envelope. We used a divide and conquer approach to break this problem into the sub problem of fitting a single to the source. We used the standard optimization method SQP to solve this sub problem

Вам также может понравиться