1. Performance of Service- Node-Based Mobile Prepaid Service Ming-Feng Chang, Wei-Zu Yang, and Yi-Bing Lin, Senior Member, IEEE IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, MAY 2002

2. Outline Introduction Mobile Prepaid Service Four billing technologies used in prepaid service Service node approach Analytic model Numeric examples Conclusion

3. Mobile Prepaid Service Prospect of the mobile prepaid service USA, 1998 grew 56% to 2 billion US dollars.will maintain a high growth rate to 2005 Taiwan, FarEastone, more than 40% of their customer subscribed to prepaid service What is the prepaid service Prepaid card includes an directory number and the credit Charge is decremented from the remaining prepaid credit Whisper tone reminding this person to recharge When zero balance, cannot originate calls, but may be allowed to receive phone calls for a period top-up card(scratch card with a secret code inside it.) to recharge the prepaid credit

4. Mobile Prepaid Service Customer ’ s point of view immediate service without a long-term contract or regular bills System provider ’ s point of view reduces operation overhead reduces the time of cost reclamation Increases the capability of competition

5. Mobile Prepaid Service Provider ’ s interest Cost v.s. revenue # of credit checks v.s. bed debt Four billing technologies used in prepaid service Hot billing approach handset-based approach intelligent network approach service node approach

6. billing technologies(1/4) Hot billing approach uses call detail records (CDRs) produced by the wireless switch (i.e., mobile switching center) to process the prepaid usage CDRs generated after call completionsan,transported from the mobile switching center(MSC) to the prepaid service center. cost effective,not require major changes in the network infrastructure One call exposure problem,large loss, bed bedt

7. billing technologies(2/4) handset-based approach prepaid credit and balance information is stored in the SIM card of a handset.. the MSC provides tariff parameters to the handset through the GSM phase 2 supplementary message AOC (advice of charge) Handset modify the balance information in the SIM card not incur major modification to the carrier’s infrastructure. requires GSM phase-II-compliant handset security is a serious problem, network needs to act as a backup to keep track of the prepaid credit usage.

8. billing technologies(3/4) intelligent network approach Complete solution for the prepaid service service control(checking credit, timer…),not on MSCs, but on the prepaid service control point (P-SCP). P-SCP contains service logic programs (SLPs) and associated data to provide IN services prepaid call to MSC, MSC communicates with the P-SCP, P-SCP performs the service control and response message back to the MSC, MSC accept or reject the prepaid call P-SCP is not on the voice path, low capacity expansion cost. Investment on P-SCP and software modifications in all MCSs

9. billing technologies(4/4) Service node approach(the paper ’ s focus) most widely deployed prepaid solution today viewed as a stepping-stone to the intelligent network approach integrates the functions of the MSC and service control point (SCP) in a closed configuration service node is on the voice path, capacity expansion cost is higher than the intelligent network approach. No need of software modifications in MCSs

10. Service node approach 1 originates a prepaid call by dialing 1 2 routes the call to the service node 3 verify if the customer has sufficient credit service node activates a timer for charging and sets up a trunk back to the MSC 4 4 Route to destination 4 Negative credit !!! Terminate by the service node. IVR may be reminding 5 5 5 Call completes, credit updated 6

11. Service node approach In real operation, a service node may process over 10 000 prepaid calls simultaneously. service node need to permit a real-time credit monitoring and updating the processing budget for a service node should be accurately planned in real world “What is the credit checking frequency so that the sum of the credit checking cost and the bad debt is minimized?” Min { checking cost + bad debt }

12. Analytic model Assumption : a customer will consume all the prepaid credit before he/she gives up the prepaid service E[N * ch ] : expected number of credit checks E[B * L ] : expected bad debt B : prepaid credit K : number of calls I : decremented amount periodically by service node during the conversation X i : the charge of the ith call, i = 1,2,…, k-1 X k : the charge of the last call if the service node would not terminate the call when the credit becomes negative B * L : the loss of the service provider B L : the corresponding value if the last call were allowed to complete E[N ch ] : expected number of credit checks assuming the total credit is B + B * L E[n ch ] : the expected number of credit checks for a call P : recharge probability

13. Analytic model The last (i.e., the kth) call terminates in one of the two cases: the last call is forced to terminate by the service node the last call completes before the service node discovers that the credit becomes negative. E[N * ch ] = E[N ch ] - E[B L ]/I = E[K]E[n ch ] - E[B L ]/I

14. Analytic model Derivation of E[N * ch ] and E[B * L ], consider two cases for prepaid credit B Fixed Credit Case

15. Analytic model Recharged Credit Case

16. Numeric examples the analytic results are consistent with the simulation results

17. Numeric examples Effects of the Variation of Call Charges C x : the coefficient of the variation of call charge A large C x represents that there are more short calls and long calls. For Cx < 5*10^-3, E[N* ch ] and E[B* L ] are sensitive to E[x i ] But insensitive to Cx When Cx > 5, E[N* ch ] increases sharply in both fixed credit And recharged credit cases---short call effect.  billing policies to discourage short calls

18. Numeric examples Effect of I on E[B * L ]/I Intuition : E[B * L ] = I/2, but wrong (Obs 1)Negative slope : as I increases, the probability that the Kth call terminates normally increases. Thus, the expected loss E[B * L ] becomes smaller than I/2. (Obs 2) irregular pattern: When Cx is large, there are more short calls and long calls, the probability that the last call depletes all or most of the credit becomes large

19. Numeric examples The Cost Function C = E[B * L ] + Ø E[N * ch ] Ø : the credit checking cost of the service node. C : the net effect of credit checking cost and bad debt. Intuitive results: 1. as Ø increase, the value for the optimal I (triangle) increae 2. For the same Ø,as B increases, the value of optimal I increases. Triangle:the cost for the optimal I

20. Conclusion study the service-node based approach system architecture The procedures for call origination. An analytical model to analyze the performance the fixed credit the recharged credit cases. Validate analytic results by simulation experiments short call effect, suggest billing policies to discourage short calls When Cx is high or I is large, E[B * L ] = I/2 A cost function was used to determine the minimal cost and the optimal checking interval I