A novel statistical and distributed CAC algorithm for IEEE 802.11 based single and multi-hop wireless ad hoc networks

Abstract

In this paper, we propose a statistical Call Admission Control (CAC) schemes for single and multi-hop IEEE 802.11 based wireless ad hoc networks. Unlike most papers which consider average end-to-end delay, our algorithm guarantees statistical delay that would be more efficient for real time multimedia applications. We use Effective Bandwidth/Effective Capacity theory to compute delay bound stochastically in each hop. The proposed distributed CAC algorithm is suitable for ad hoc networks. Also, with a few changes, it has been implemented in the AODV routing protocol. Simulation results demonstrate that proposed CAC algorithm works efficiently in both single and multi-hop networks. Moreover, we investigated our algorithm with aggregated traffic and the result shows that the algorithm performed accurately under that condition.

Appendices

Appendix 1: Proof of Proposition 1

In Eq. (10) $D_{EE}(t)$ is equal to:

$$\begin{aligned} D_{EE}(t)=D_1(t)+D_2(t)+\cdots +D_i(t)+\ldots D_H(t) \end{aligned}$$

(15)

By substituting (15) in (8) and supposing $\gamma (\lambda )\approx 1$ we have:

$$\begin{aligned} \begin{aligned}&\sup _t\,Pr{\left\{ D_i(t)>D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right\} } \\&\quad \approx e^{-{\theta _i^c\left( \lambda _i\right) *\lambda _i*\left( D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right) }} \end{aligned} \end{aligned}$$

(16)

Now if we suppose $\theta _i^c(\lambda _i)*\lambda _i$ is equal to the minimum of QoS exponent in the path, namely equal to (11), (16) is given by:

$$\begin{aligned} \begin{aligned}&\sup _t\,Pr{\left\{ D_i(t)>D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right\} } \\&\quad \approx e^{-{\theta _i^c\left( \lambda _i\right) *\lambda _i*\left( D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right) }}\\&\quad \le e^{-{\theta ^c(\lambda )*\lambda *\left( D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right) }} \end{aligned} \end{aligned}$$

(17)

In homogeneous networks we assume that the delay in all hops are the same and equal to $\frac{D_{max}}{H}$, the upper bound in multi-hop network is yield to:

$$\begin{aligned} \begin{aligned}&\sup _t\,Pr{\left\{ D_i(t)>D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right\} } \\&\quad \approx e^{-{\theta _i^c\left( \lambda _i\right) *\lambda _i* \left( D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right) }}\\&\quad \le e^{-{\theta ^c(\lambda )*\lambda *\left( D_{max}-{\sum _{j=0}^H}_{j\ne i}D_j(t)\right) }} \le e^{-{\theta ^c(\lambda )*\lambda *\frac{D_{max}}{H}}} \end{aligned} \end{aligned}$$

(18)

So the proposition is proven.

Appendix 2: Flowchart of CAC algorithm

In this appendix, the flowcharts of the implemented algorithm in NS2 are represented. Figures 15, 16 and 17 show the modified RREQ function, the implemented RREP function and the modified receive function in NS2 for our CAC algorithm, respectively (Fig. 14).

Fig. 14

Start point of implemented algorithm

Fig. 15

The modified RREQ function in NS2

Fig. 16

The modified RREP function in NS2

Fig. 17

The modified data packet function in NS2