Network Planning of the CATV communication networks Arthur, K. W. Peng, OPLAB d6725005@im.ntu.edu.tw Advisor: Frank Yeong-Sung Lin April 22, 2005 Agenda Introduction Problem Description and Formulation Formulation Analysis and Reformulation Numerical Experiments Conclusion Q&A CATV Communication Network Technology TRUNK NETWORK Satellite dish Trunk Amplif ier Head End Radio tower Bridger Amplif ier Splitter Direction Couplers Tap Distribution Network Line Extender Telev ision The Network Structure of CATV Networks Network Planning --Traditional Approaches 製圖 幹線系統設計 餽線系統設計 反向系統設計 幹線系統設計 頭端幹線系統 餽線系統設計 Figure 2-10 餽線系統設計 相對信號位準與損失的系統點 反向系統設計 Similar to the design of downlink. Noise Funneling Limit the number of branch and amplifier. Addressable Bridger Leg Switch 前向---反向放大器 CAD Tools The traditional way is calculation-intensive and repetitive. Comparison of the CAD tools[Yermolov,2000] CATV CAD: Gen Enterprise Ltd. Symplex Suite of software: SpanPro Inc. Program System:Lode Data Corporation CADIX International Inc. Cable Tools: Goldcom Inc. Feature Auto-tracking the signal quality. Helping the design to calculate the network requirement, cost, etc. The design is still depend on the experience of the network designer. CATV Network Planning Tools •Stand-alone version •Web-based version Mathematical Formulation and Network Optimization Basic ideas: formulate the network and using network optimization technique to find the optimal solution. Head End Al Gv CNR X-MOD Fl Mv l Ov Bv Fv Gv Fv Mv Av (link ) ( equipment ) CNR X-MOD CSO CTB User Performance Requirements Performance requirements in downstream CNR (Carrier to Noise Ratio) ≧43dB X-MOD (Cross Modulation ) ≦-46dB CSO (Composite Second Order) ≦-53dB CTB (Composite Triple Beat) ≦-53dB Performance Requirements (cont’d) Performance requirements in upstream Problem Formulation Problem description Given : downstream performance objectives upstream performance objectives specifications of network components cost structure of network components number and position of endusers terrain which networks will pass through and the associated cost Determine: routing allocation of network components operational parameters (e.g., gain of each amplifier) Problem Formulation Features Nonlinear problems Hard to solve directly by standard methods Some techniques needed Problem Decomposition Steiner Tree Problem Network Optimization Geometric Programming Posynomial form Gradient-based Optimization Problem Decomposition and Reformulation Part I: Steiner Tree Problems min yl Cl lL (13) v yl 1 v V (14) yl 0 or 1 l L (15) pl x p yl | W | l L lLin wW pPw (16) x p 1 w W (17) x p 0 or 1 p Pw , w W pPw Problem Decomposition and Reformulation (cont’d) User3 Head End User1 User2 Steiner vertices regular vertices Problem Decomposition and Reformulation (cont’d) Heuristic approximation algorithms Minimum Cost Paths Heuristic (MPH) G (V , E , d ) S V S {v1 , v2 , v3 ,..., vk } PATH (W , s ) : shortest path from a connected component W to a vertex s in G ( PATH (W , s )) : the cost of Path(W,s) step 1 : V1 {v1} step 2 : for each i 2 ,3,...,k do find a vertex vi in S-Vi-1 such that PATH(Vi-1 ,vi )) min{ ( PATH(Vi-1 ,v j )) | v j S-Vi-1} Vi add PATH(Vi-1 ,vi ) to Vi-1 Problem Decomposition and Reformulation (cont’d) Problem Decomposition and Reformulation (cont’d) Part II min [d1 ( Al )1 ] {z v [d 2 ( Fv ) 1 d 3 (Gv )1 d 4 ( M v )1 lL vV 1 1 1 1 1 d 5 ( Bv ) d 6 (Ov ) ] z v [d 7 ( Fv ) d 8 (Gv ) d 9 ( M v ) ] z v [d10 ( Av )1 ]} L : the set of links in the given candidate topology V : the set of nodes in the given candidate topology s.t. Problem Decomposition and Reformulation (cont’d) H H pc pc (1) Si Gpi Api Apjpj 10 i 1 j 1 pn H pc z (2) ( n 1 H pc n 1 n 1 i 1 j 1 n H pc pi pi (4) z ( Bpi ) i 1 H pc ) 10 59 Csys 10 S Gpi Api Apjpj * 10 10 (3) z ( M pi ) i 1 Fpn w W 0.5 0.5 n 1 Gpi Api Apjpj 10 i 1 j 1 n n 1 Gpi Api Apjpj 10 i 1 j 1 n n 1 i 1 j 1 w W M sys 10 Bsys 10 *0.5 *0.5 Osys 10 W (5) z (Opi ) Gpi Api Apjpj 10w i 1 pi 1 Problem Decomposition and Reformulation (cont’d) H pc (6) Gpi Api Apj 1 H pc i H pv j H pv w W v V p Csys 1 5.9 0.5 (7) s 10 Ft ( z v z v ) 10 10 w W vV p vVP V p : the node set of path p VP : the node set of path set P Decision Variables : Gv : gain of upstream amplifier s : input signal strength to upstream amplifier Ft : noise figure of upstream amplifier Problem Decomposition and Reformulation (cont’d) (8) zpi ( M t ) 0.5 s Gpi 10 H pc M sys 10 *0.5 i 1 Decision Variable : M t : cross modulation of upstream amplifier w W Solution Approaches Posynomial problem min g 0 (t ) s.t. : t1 0 , t 2 0 , ... , t m 0 g1 (t ) 1 , g 2 (t ) 1 , ... , g p (t ) 1 g k (t ) c t iJ [ k ] ai 1 ai 2 i 2 2 t ...t maim , k 0, 1, ..., p, aij : arbitrary real numbers ci : positive gk(t) : posynomials (IP) (1) (2) Solution Approaches (cont’d) Dual problem n max v ( ) [ ( i 1 ci i p ) i ] k ( ) k ( ) k 1 s.t. : 1 0, 2 0, ... n 0 i 1 a ij i 0 i 1 k ( ) i , iJ [ k ] Positivity condition Normality condition jJ [ 0 ] n (IP) j 1,2,..., p k 1,2,..., p Orthogonality condition Solution Approaches (cont’d) Penalty method min n ln v( ) J 1 ( i 1) J 2 ( aij i ) 2 j 1,2,..., m 2 iJ [ 0 ] i 1 where J 1 and J 2 are large positive numbers. Steepest descent method Rounding procedure Computational Experiments Solution modules Module 1 Determining the Interconnection and Routing of CATV Networks Module 2 Determining Locations to Place Amplifiers Module 3 Determining Configurations and Parameters of CATV Components Module 4 Determining Configurations and Parameters of CATV Reverse Modules Computational Experiments 91 100 90 81 71 80 61 70 51 60 41 50 31 40 21 30 11 20 1 2 3 4 5 6 7 8 9 10 Computational Experiments The constructed steiner tree 91 100 90 81 71 80 61 70 51 60 41 50 31 40 21 30 11 20 1 2 3 4 5 6 7 8 9 10 Computational Experiments Deciding the locations and parameters of amplifiers 91 100 zg=0.053235 81 90 71 zg=0.064211 80 61 70 51 60 41 50 31 zg=0.064212 21 30 11 1 40 20 zg=0.117608 2 3 4 5 6 7 8 9 10 Conclusion It is feasible to use mathematical programming methods in CATV network planning The solution provided by this approach can be used to evaluate the QoS in many situation. As a core module, we can add more features: New network components New Services Future Research Directions Network Planning and Management CATV network planning and optimization Layering QoS Fault tolerance/Reliability CATV network performance Capacity management Admission Control Q&A