Abstract

Capacity-on-demand is the key concept in multiplexing bursty mobile data traffic over wireless links featuring limited bandwidth. This scheme maintains a connection for a mobile only when it has data to transfer and allows quick release of radio resource when buffered data is emptied. For a system supporting a large number of data mobiles, non-trivial multiple access delay may be introduced and its distribution may very likely follow an arbitrarily general form given the complicated multiple access scenario. A general mobile data terminal model, namely MMPP/G/1/L with server activation time, is presented in which we consider bursty traffic input, finite buffer, and generally distributed connection activation time and service time. Performance indices are derived from the model and can be used in the performance prediction on higher layers and thus to help predict user satisfaction for potential data services. In particular, we use the model to study how performance will be affected by the distribution of access delay in capacity-on-demand.

Author Keywords: Author Keywords: Capacity-on-demand; Quality of service (QoS); Burstiness; Markov regenerative process (MRGP); Uniformization

Article Outline

1. Introduction

2. The mobile model

2.1. Model description

2.2. Model analysis

2.3. Steady-state solution

2.4. Performance measures

3. Numerical results and discussion

3.1. Activation time distribution and performance

3.2. Traffic pattern and performance

4. Conclusion

Appendix A

A.1. Markov regenerative processes [8]

A.2. Matrices K(t) and E(t)

A.3. Computing I₁^F(k;n) and I₂^F(k;n)

References

Vitae

Corresponding author. Tel.: +1-9194612445x225.