1 / 63

- Fei Yu, Vincent W. S. Wong, Victor C. M. Leung Presented by Kiran Kumar Bankupally

Performance Enhancement of Combining QoS Provisioning and Location Management in Wireless Cellular Networks. - Fei Yu, Vincent W. S. Wong, Victor C. M. Leung Presented by Kiran Kumar Bankupally. What is this paper about?.

tobit
Télécharger la présentation

- Fei Yu, Vincent W. S. Wong, Victor C. M. Leung Presented by Kiran Kumar Bankupally

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Performance Enhancement of Combining QoSProvisioning and Location Management inWireless Cellular Networks -Fei Yu, Vincent W. S. Wong, Victor C. M. Leung Presented by Kiran Kumar Bankupally

  2. What is this paper about? • Framework for Integrating QOS provisioning and Location Management • Proposing a new Connection Access Control (CAC) using the integrated Scheme.

  3. Key Words • MS:- mobile station • CAC:- Connection Admission Control • LM:- Location Management

  4. Introduction • Tracking user is an issue.(Location Management) • Requirement of diverse QOS requirements • Due to mobility availability of resources at connection setup doesn’t guarantee the resource availability. • Performance degradation due to mobile hand offs.

  5. Location Management • Divided into two parts • Paging • Location Updating • QOS problems • Deal with handoffs • Forced connection terminations due to handoff blocking are generally more objectionable compared to new connection blocking • Maximize utilization to reduce handoffs of new connections

  6. Key motivations • User mobility is the main problem for both LM and why CAC is required, but all the previous works deal with them on different sets of information. • per-user mobility pattern can provide the basis for effective solutions that address these two sets of system requirements, it will be helpful to consider them jointly and make them share information with each other. • both in-session and out-of-session movements are parts of a user’s mobility pattern and using all available mobility information will improve the performance of both CAC and LM schemes.

  7. Proposed Framework

  8. Contribution • Contribution of this work is three fold • more efficient and cost effective solutions because of the integration • New path based LM scheme which uses all the available mobility information in both location update and paging processes • Novel CAC scheme is proposed. Predict not only where the MS will move, but also when the MS will move to a new cell based on the mobility prediction.

  9. COMMON MOBILITY PREDICTION SCHEME • purpose?? • For Location management • Minimizing paging cost • For QOS provisioning • Better CAC Design • Rationale?? • Most users have a favorite path which they repeat most of the times. • Shows the stationarity of sequence symbols . • Motivation?? • Optimal Data Compression methods

  10. Model Assumptions • Network Topology • Previous work used hexagonal of Square lattices for the cell arrangement which is improper because of antenna radiation pattern and propagation environment. • Graph model is used here. network is modeled as connected Graph G = <V,E> V = set of base stations each representing a single cell E = adjacency between cells.

  11. Example topology

  12. Model Assumptions(Cont..) • Channel Holding Time and Cell Residence Time • Definitions • Channel Holding Time:- time during which a connection occupies a channel in a given cell • Cell Residence Time:- amount of time that the mobile user stays in that cell • Previous models assume them to be exponentially distributed, independently and identically distributed for all cells. Here they are assumed to follow General distributions

  13. Model Assumptions(Cont..) • User Mobility Model • Symmetric random walk model was used previously which ignores the favorite path concept. It assumed equal probability to all neighboring cells • Here, MSs future locations are predicted by correlating with its movement history

  14. Prediction overview • Let In a cellular network, the mobility of a user can be represented by a sequence of symbols,C1,C2,C3.. , where Ci denotes the identity of the cell visited by the MS • the sequence of symbols is assumed to be generated by an mth order Markov source, where the states correspond to the contexts of the previous symbols • Probability depends on the current cell or a list recently visited cells.

  15. The Optimal Data Compression Algorithm • a dictionary-based compression algorithm that performs incremental parsing of an input sequence which is optimal theoritically and good practically • This algorithm parses input string x1, x2, ..xi into c(i) substrings w1, w2, ..wc(i) such that for all j>0, the prefix of the substring wjis equal to some wi for 1<i<j • Uses a Trie that feeds the probability information to arithematic encoder which encodes a sequence of probability of p using –log2(p)

  16. Example • The Alphabets are {a,b,c} • Input string • ababcabcababcabc…. • (a)(b)(ab)(c),(abc),(aba),(bc),(abc..)… • For this example the encoder since we are analyzing 7 values the arithematic encoder encodes the sequence with log27 b

  17. Trie Construction Root a,1 • Step:-(a)babcabcababcabc….

  18. Trie Construction Root a,2 b,1 • Step:-(a) (b)abcabcababcabc….

  19. Trie Construction Root a,2 b,1 b,1 • Step:-(a) (b)(ab)cabcababcabc….

  20. Trie Construction Root c,1 a,2 b,1 b,1 • Step:-(a) (b)(ab)(c)abcababcabc….

  21. Trie Construction Root c,1 a,3 b,1 b,2 c,1 • Step:-(a) (b)(ab)(c)(abc)ababcabc….

  22. Trie Construction Root c,1 a,4 b,1 b,3 a,1 c,1 • Step:-(a) (b)(ab)(c)(abc)(aba)bcabc….

  23. Trie Construction Root c,1 a,4 b,2 b,3 c,1 a,1 c,1 • Step:-(a) (b)(ab)(c)(abc)(aba)(bc)abc….

  24. Mobility Prediction Scheme • Is similar to prediction by partial matching(PPM) data compression algorithm • PPM algorithm • Basis of PPM of order m is a set of m+1 Markov predictors. • A Markov predictor of order j predicts the next event based on the j immediately preceding events • A trie is used to store all m contexts called mobility trie

  25. Pseudo Code for Mobility Prediction Scheme

  26. Example (2) • Input sequence:-ababcabcababcab… • The Trie(This is not Le-Zi but Active Le-Zi)

  27. Example (Cont..) • Scenario:- last three cells visited is abc. • Want to predict next cell th MS will visit. • Method:- • First estimate the probability distributions for 0,1 and 2. P2a=1 P2b=0 P2c=0 P1a=1 P1b=0 P1c=0 P0a=5/13 P2b=5/13 P2c=3/13

  28. Example (Cont..) • Blending vectors {w0,w1,w2} where • The weights can be fixed or adapt as prediction proceeds to give more emphasis to higher models. • Then the probabilites assignment is given by

  29. Implementation issues • Deciding a data structure to store a trie is important. • have a pointer structure similar to trie structure • a linked list implementation • Hashing can also be used. • Also to reduce memory and computation complexity size of data structure is limited. • Explicit bound to M. • LRU strategy

  30. Pointer Structure

  31. Linked List structure

  32. LM in combined Fraework • A path based approach is used with slight changes. • All available location information is used in prediction. • In original one location during the connection is treated same as a normal one. • Here during session the Location update is done.And when out of session then wait for new pattern.If CMR is high you always have the track • In session LU doesn’t need much resources.

  33. Example

  34. Numerical Results • Simulation Environment Features • Number of base stations 40 • Average number of neighbors 6 • Each mobile user has 5 different paths( since every one has a favorite path) with probabilities 0.6,0.2,0.1,0.05,0.05. • Paths are generated by first selections two nodes at random as origin and destination nodes and whenever the mobile user leraves a cell it moves to a neighbouring cell which is closest to destination

  35. Numerical Results(Cont…) • Cell residence Time follows i.i.d Gamma distribution with avgtim 1/μrasds • New connection arrival time λ per minute. • Connection durations are exponentially distributed with mean 1/μd which is 3 min. • ρ = λ/μrCMR • Location Update is done using movement based scheme.here after every movement of 1 cell a update is done for simplicity.

  36. Numerical Results(Cont…) • For comparison purposes update(original) represents number of update messages in original scheme and similarly for update(new) • Same goes with paging(original) and paging (new) • Performance gain in updates • PG = update(original)/ update(new) • Performance gain in paging • PG = paging (original)/paging (new)

  37. Update Gain

  38. Paging Gain

  39. CAC in new framework • Terms used • CAC:- Connection Admission Control • Phd :- probability if handoff connection being dropped • Pnb :- probability new Connection is dropped • E-OTD:- Enhanced Observed Time Difference technology • BTS:- Base transiever Stations • MS :- Mobile Station

  40. CAC in new framework • Due to in session user mobility CAC needs to perform mobility related QOS provisioning in cellular networks. • Key idea is to predict next node MS visits and try to acquire the resource before hand considering time it takes to reach that node. • If resource is available, it is reserved for MS to guarantee Phd • E-OTD technology is used in this scheme.

  41. E-OTD and Time Interval Prediction • Time Interval Prediction is done using this. • Unknown MS position p =(x,y) is estimated by using the Time Difference Of Arrival(TDOA) measurements between the MS and known Coordinates,BTSsof known coordinates. • TDOA Property • TDOA between BTS1(serving BTS) and BTSi(i=2..N-neighbouring BTS)defines a hyperbola whose focii coincide with coordinates of two BTSs. • Two Hyprebolas are minimum to estimate MS position.

  42. E-OTD and Time Interval Prediction(Cont…) • TDOA is defined as Geometric Time Difference (GTD) where tRxi and tTxi are, respectively, the reception and transmission epochs of the burst from the ith BTS.

  43. E-OTD and Time Interval Prediction(Cont…) • Further simplifying where OTD = Observed Time Difference RTD = Real Time Difference In absence of errors the position can be accurately measured by

  44. E-OTD and Time Interval Prediction(Cont…) • Because of noise Eq1 doesn’t hold good. • linear regress setup can be used to smooth the data for more accurate velocity and position estimation of an MS • K- previous estimations are used to obtain the MSs current estimated velocity and position • represents estimated locations at subsequent time points tn.

  45. E-OTD and Time Interval Prediction(Cont…) • Velovity is given by

  46. E-OTD and Time Interval Prediction(Cont…)

  47. E-OTD and Time Interval Prediction(Cont…) • Let ta(i,j) denote the time when MS in cell i will arrive at cell j, and td(i,j)denote the time when the MS in cell i will depart from cell j. The values can be calculated as • Where d(p(t),j) is the distance between the current position p(t) and the boundary of cell i and j, and d(j) is the route distance inside cell j.

  48. CAC scheme • Idea:- Verify the feasibility of accepting new and handoff connections under the conditions of guaranteeing the QOS of existing connections and maximizing the utilization • Achieved by the predictions of where an MS will visit using the scheme and when an MS will visit • P(i,j,ta,td)the probability that an MS original in cell i will visit cell j during the time interval taand td.

  49. CAC Scheme(Cont..) • Connection duration follows Exponential distribution with rate ud. • Where P(i,j) is calculated from Trie.

  50. CAC Scheme(Cont..) • When an MS is active in cell ,we can obtain the most likely cell-time (MLCT) of that MS, a cluster of cells and time where and when the MS will most likely visit in the future. • MLCT is defined as • Required bandwidth to be reserved in cell j for the expected handoff of m from cell i

More Related