A New Approach to Modeling Time Dependent Problems in Wireless Broadband Networks
| Autor: | Vijayalakshmi Chetlapalli, Himanshu Agrawal, K. S. S. Iyer, Sunil Patil |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
021103 operations research
Stochastic process Real-time computing 0211 other engineering and technologies 02 engineering and technology Interval (mathematics) Expected value Poisson distribution 01 natural sciences Point process 010309 optics symbols.namesake Offered load Base station Wireless broadband 0103 physical sciences symbols Mathematics |
| Zdroj: | Proceedings of the ACM Workshop on Distributed Information Processing in Wireless Networks. |
| DOI: | 10.1145/3083181.3083182 |
| Popis: | At the base station of a wireless broadband network, connection requests for various services viz., voice, browsing, streaming video, online gaming etc., arrive at random points of time and depart at random points of time. The arrival rates of connection requests for various services change with time of the day, resulting in non-stationary aggregate traffic load that is mathematically complex to model. We propose a novel approach, by using two techniques of stochastic point processes viz., Product Density and Random Point Processes to model time dependent problems in wireless broadband networks. We classify the problems faced by a wireless broadband service provider into two categories---counting measures and aggregate measures. By counting measures, we mean expected number and fluctuation in (a) number of calls of each traffic type in service at any time of the day (b) number of calls that have completed service in a time interval (c) number of calls that have exceeded a given delay threshold and (d) offered load. We define Product Densities of first order to evaluate the expected value and Product Densities of second order to evaluate the fluctuation of these measures for non-homogeneous Poisson arrival rate and exponential servicing rate. Each of the connection requests arriving at random instants of time triggers a random process of traffic, for a random duration. The cumulative built-up of the triggered random process is non-Markovian and hence, it is mathematically complex to estimate its statistical characteristics. We categorize as aggregate measures, the expected value and fluctuation in (a) aggregate traffic load at the base station from all types of traffic and (b) aggregate resources (or bandwidth) required to service this traffic load. The cumulative built-up of these measures fits in with Random Point Process, a special class of stochastic point processes. We estimate the expected value and fluctuation of aggregate measures for time dependent arrival rates, using Random Point Process and Product Density. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|