Query Processing in a Mobile Computing Environment Exploiting the features of Asymmetry

1. Query Processing in a Mobile Computing EnvironmentExploiting the features of Asymmetry By,Wen-Chih Peng, Member, IEEE&Ming-Syan Chen,Fellow,IEEEPresented By, Sandeep Bomidika

2. ABSTRACT This paper describes about the concepts of • Mobile Database • Mobile Computing • Query Processing • Join Method

3. Introduction: In a mobile computing environment, a mobile user with a power-limited palm computer (or a mobile computer) can access various information via wireless communication. Applications include • stock activities • traffic reports and • weather forecasts

4. The query processing in a traditional distributed system is composed of the following three phases: 1. local processing phase 2. reduction phase and 3. final processing phase.

5. ASYMMETRIC FEATURES The three asymmetric features, which we shall explicitly address and reflect in the design of query processing schemes, are as follows: 1. Asymmetric feature of computing capability between the server and a mobile computer.2. Asymmetric feature of energy consumption betweenmessage sending and receiving.3. Asymmetric feature of energy consumption betweenactiveness and idleness of a mobile computer.

6. QUERY SCENARIO OF JOIN PROCESSING A query scenario of join processing.

7. PRELIMINARIES: 1. NOTATION , DEFINITION AND ASSUMPTION • Query processing in a mobile environment, which involves join processing among different sites including static servers and mobile computers. • A mobile computing system consists of stationary servers and mobile computers. Stationary servers include information servers and the equipment of a mobile communication system.

8. Example: TABLE 1 An Illustrative Example for Semijoin Operation

9. 2. COST MODEL FOR JOIN AND QUERY PROCESSING IN A MOBILE COMPUTING SYSTEM • The processing time of joining Ri and Rj at a mobile computer can be expressed by rSM* TS (Ri l><l Rj). • The amount of energy consumed in processing the join between Ri and Rj at a mobile computer is devised as 1/δ* rSM * TS(Ri l><l Rj).

10. JOIN PROCESSING IN A MOBILE COMPUTING SYSTEM • Processing the Join at the Server (Denoted by Js) • Processing the Join at the Mobile Computer (Denoted by Jc) • Employing an MI Profitable Semijoin for Join Processing (Denoted by Jsc)

11. Analysis of Join Processing • Lemma 1.The amount of data transmission incurred by join method Jc is smaller than that by join method Js ,i.e , dt(Jc) < dt(Js). • Lemma 2. With a given selectivity factor ρ2,a, where ρ2,a < |R1| |A|+|R1| , the amount of data transmission incurred by method JSC is smaller than that by method Jc; i.e., dt(Jsc) < dt(Jc). • Lemma 3. Ec(Jc) > Ec(Js). A semijoin is called MI profitable if and only if Ec(Jsc) is smaller than Ec(Js).

12. QUERY PROCESSING IN A MOBILE COMPUTINGSYSTEM • Query Processing at the Destination Mobile Computer (Denoted by QPc). • Query Processing at the Server(Denoted by QPs) • Query Processing with SI Profitable Semijoins (Denoted by QPsj)

13. Algorithm: Algorithm M.Determine the join sequence with SI profitable semijoins interleaved in the RT phase. Input: A directed graph = (V-d , E-Ed). Output: SEQ /* SEQ contains the resulting sequence of joins semijoins */ 1. begin 2. SEQ = Ø; 3. for each vertex w Є V - d do 4. begin 5. w. mark := false; /* w. mark is used to indicate if w has been visited or not */ 6. w. ct := ∞; /* w. ct is the cost of join operation from S to w. */ 7. w. op = J; /* w. op represents the join operation */ 8. end 9. S. ct = 0; /* The cost of join operation from S to itself is set to zero */

14. 10. while З an unmarked vertex do 11. begin 12. let w be an unmarked vertex such that w.ct is the minimal among all the corresponding costs of unmarked vertices; 13. if w.ρ < (rE *er-1/δ * rSM*tuple) (rE *er+1/δ * rSM*tuple) /* Determine if a semijoin is SI profitable */ 14. w:op = SJ; 15. SEQ=SEQ U w; 16. w:mark := true; 17. for all edges (w; z) with z is unmarked do 18. begin 19 . if w:ct+weight(w; z) < z:ct then 20. z:ct :¼ w:ct+weight(w; z); /* Update the weight of the edge (w,z) */ 21 . end 22 . end 23. end

15. A Profile for a Query Where |A|= 19; |B| = 15;|C| = 17; |D| = 19; |E| = 16; |F| =15; |G| = 18;rSM= 5;δ = 0:5; er = 0:1; rE = 5; and ttuple = 0:01

16. A Directed graph for a query graph

17. Execution scenario of algorithm M

18. An Execution example of Algorithm M

19. Parameters Used in Simulation

20. Performance Evaluation • System Model

21. Experimental Results of Join Processing • Experiments of Selectivity Factor

22. Characteristic Function of MI Profitable Semijoin for Selectivity Factor

23. Experiments of Idling Coefficient The energy consumptions of Js, Jc, and Jsc with δ varied.

24. Characteristic Function of MI Profitable Semijoinfor Idling Coefficient

25. Experimental Results of Query Processing: • Experiments of SI Profitable Semijoin The performance of SI semijoins and simple joins with rsm = 5. (a) The energy consumption of SI semijoin and simple join. (b) The amount of data transmission of SI semijoin and simple join.

26. Characteristic Function of SI Profitable Semijoin The characteristic function of an SI profitable, fsi (.), with rSM varied

27. Performance of QPs and QPsj The performance of QPC, QPS, and QPSC when the number of relations is 5.

28. The amounts of data transmission and energy consumption incurred by QPS and QPSJ in the RT phase with the number of relations varied.

29. Conclusion: • Explored three asymmetric features of a mobile environment • Devised query processing methods for both join and query processing. • Examined three different join methods and devised some specific criteria to identify MI/SI profitable semijoins. • By exploiting the three asymmetric features, these characteristic functions are very powerful in reducing both the amounts of energy consumption and data transmission incurred and can lead to the design of an efficient and effective query processing procedure for a mobile computing environment.

30. THANK YOU