Journal of Electronics and Communication Engineering & Technology (JECET)

Transcription

1 Journal of Electronics and Communication Engineering & Technology (JECET) JECET I A E M E Journal of Electronics and Communication Engineering & Technology (JECET)ISSN ISSN (Print) ISSN X (Online) Volume 1, Issue 1, July-December (2013), pp IAEME: CLUSTERED CONDUCTION OF VOIP ROUTING TOPOLOGY FOR WLAN Mohammed Sirajuddin 1, Dr D. Rajya Lakshmi 2 and Dr Syed Abdul Sattar 3 1 Royal Institute of Technology and Science, Chevella, Hyderabad, India 2 Gitam Institute of Technology, Gitam University, Vizag, India 3 Royal Institute of Technology and Science, Chevella, Hyderabad, India ABSTRACT In this Paper, we propose a routing topology for VoIP transmission over WLANs that referred as clustered conduction of VoIP routing (CCVR) topology, which is Quality of Service Centric. A novel scheduling mechanism introduced to differentiate the capricious loss of the data packet and transmission delay. The aim of the proposal is to achieve a clustered approach in the tasks involved in VoIP routing. The expected clustered approach splits the routing functionalities such as scheduling and buffering process. Keyword: , WLAN, QoS, ad hoc network, CCVR 1. INTRODUCTION Real-time VoIP systems are suitably more and more popular in a range of applications. The broad choice of applications embraces group collaboration, remote medical diagnosis/treatment, conferencing systems, on-demand video services, and distance/remote sensing and monitoring [1]. It is clear that the Internet is integrating a Global Mobile system and different Clusters of systems into a big standard IP network [2]. IP based WLAN to be our technology to take VoIP applications. We chose a micro-mobility topology as a solution to the performance and scalability issues of WLAN [3].VoIP applications have very unusual necessities from applications for which the internet was initially designed. VoIP applications are real-time applications that need a certain amount of bandwidth to make sure the bit-rate wanted by each media stream and harsh delay difference wants to keep away from buffer underflow at the receivers. We require a mechanism to monitor the network nonintrusively to notice any change. Moreover, for supporting VoIP streams, a definite means for supporting mixed quality of service (QoS) required. The paper is organized as follows: Section 2 briefly describes the related work. Section 3 describes the proposed clustered conduction and ordering of services. Section 4 describes the Simulation and performance analysis and Section 5 concludes the paper. 27

2 2. RELATED WORK Many researchers have addressed the issue of supporting QoS for IP based WLAN by using scheduling algorithms. Simple scheduling algorithms such as First- Come- First-Served (FCFS) were first recommended. In [4], the early deadline-first (EDF) algorithm is extensive for scheduling real-time traffic in an IP-based WLAN network. An algorithm called TBLB (Token Bank Leaky Bucket) [5], which is using admission control and device policing to provide real-time VoIP traffic streams in packet-switched networks [5]. The drawback of this approach is that the modeling overhead rise linearly with the number of flows present in the network.fair queuing proposed approach that is specific to real-time traffic [6]. Increasing the weight for specific Clusters may usually result in better performance with respect to delay. However, it is a complex task to find suitable values for the weight, even in an ideal generalized processor sharing (GPS) scheduler. Recently, it seems to have been agreed upon that type specific architectures will be a viable solution for providing service guarantees in the Internet. Compared with stream specific architectures, since type specific architectures are working with simpler algorithms for implementing QoS guarantees, and hence, they can be deployed with minor changes to the network architecture. 3. CLUSTERED CONDUCTION AND ORDERING OF SERVICES The proposed clustered approach buffers the VoIP data by Clustering that data into different clusters based on the data properties and for each Cluster a different buffer cluster will be allotted. The model we use relies on using adaptive real-time scheduling method that takes into account the characteristics of the wireless channel as well as the QoS requirements for each traffic flow. The diverse parameters are considered in developing this model and their description follows. Each cluster of traffic is connected to the relevant cluster of the buffer that buffers the packets in the form of FIFO, which can be referred as packet clustering. The packet that streams into a network through a gateway router will be moved to the relevant buffer cluster that selected from a cluster to which that packet belongs to. The packet clustering process not aware of the stream to which that ingress packet belongs to. In clustered buffering each cluster is having a finite size, which can be differ from the sizes of other clusters, but the total size of all clusters together must be equal to the actual buffer. This can be referred as Clustered Inflow Buffering Upon an ingress packet moved to its buffer cluster, the service priority ranks will be normalized according to the load at individual buffer clusters in order to achieve the service agreement. In this regard, often packets can be discarded from some of buffer clusters. This can be referred as Clustered egressing As a matter of generalizing, enhanced delay tolerance and utmost packet drop possibility and ability of detecting fair changes in target network can be considered as QoS aspects for VoIP data transmission. Service priority ranking under service scheduler must be done under QoS aspects. This process can be elaborated with an example that follows let us consider VoIP traffic at a gateway router. Here the ingress is Clusterified into three clusters and the same number of 28

3 buffer clusters are used to store these three Cluster Clusters of ingress. If the clusters are being prioritized in the same order such that Cluster one get top priority, cluster 2 is in second order and cluster 3 follows cluster 1 and cluster 2 with rank 3. Then few QoS aspects can be defined as follows: The delay in cluster 2 data must be double to the delay at cluster 1 data: This QoS aspect helps to achieve the better throughput at less Frame Overhead. This QoS aspect can be generalized as delay in a cluster A data must be double to the delay in a cluster B data, if cluster A rank follows cluster B rank in order. The packet loss in cluster 3 must be equal to packet loss in cluster 2: This QoS aspect helps to achieve the better bandwidth utilization with fewer Frame Overhead. This QoS aspect can be generalized such that the packet loss ratio of two clusters that are sequenced in given priority ranks must be same. The enhanced delay of packets at cluster 3 must be less than the max delay threshold: This QoS aspect fixes the finite state for delay enhanced by a buffer cluster during the transmission; hence the packets that can t survive beyond the max delay time can be dropped from the buffer to accommodate for other capable packets in the stream. This QOS aspect helps to achieve the effective buffer management. This QoS aspect can be generalized such that the max delay threshold of the cluster that stands lost in the priority rank order must set to finite state. Because we desire to maintain complete guarantees and do not use admission control, a set of service guarantees may be infeasible in some occasions. As an instance, achieving delay bound and loss rate bound together at burst of traffic is not practically possible. If inconsistency observed in servicee guarantees then some of the service guarantees need to be tranquiled. For instance, to achieve absolute bounds the proportional guarantees need to be relaxed, and the same way to achieve the delay guarantees loss guarantees need to be relaxed. In the assumption of QoS guarantees with precedence order that uses to conclude the constraints to be relaxed against to inconsistent service guarantees. Figure 1: Architecture of clustered transmission in wireless WLAN networks 29

4 There are two requirements that a relative separation model should meet: Cluster space balancing is the network operator should be able to regulate the cluster spacing between clusters based on their criteria. Autonomous Cluster Ordering is the separation is consistent (a higher cluster is better or at least no worse than a lower cluster) and the relative ordering between clusters should be met autonomous of load condition and time scales. The scheduling algorithm functions as follows. For each event of advent, the service rate allotment of traffic clusters is modified so that all QoS service guarantees are met. If there exists no possible rate allocation that meets all services guarantee then transmission losses due to traffic drop either from a new advent or from the current backlog. In the model we propose, the number of the queues is identical to the number of the separation clusters. The state history is maintained at a clustering stage, not on a flow state as with others future separation algorithms. Thus, the state record of a Cluster then is to consider to the transmission of a packet from a Cluster or falling of a packet from a Cluster. For Cluster priority, our scheduler will grantee Cluster C will obtain improved or at least no worse service than Cluster C i SIMULATION AND RESULTS ANALYSIS i We present an estimate of the model discussed in section (3) using NS-2 network simulator. Our aims are (1) to conclude if and how well the preferred QoS is can be achieved by differentiation; (2) that the scheduling algorithm can reach controllable and predictable interruption and packet loss separation; and (3) to compare our algorithm with existing models. We test the algorithm for combining TCP and UDP traffic; we also want to study the level of quality of service and end-to-end flow that can be received with the proposed per-node guarantees. The relationship between the CIP nodes is 45 Mbps, and sources relate to the Gateway by 126 Mbps links. Each 40 to 60 Mbps link has a dissemination delay of 3 to 4 ms; each 126 Mbps association has a dissemination delay of 1 ms. Cluster 1, 2 and 3 only consist of TCP transmission, and the 4th cluster only consists of UDP transmission flow. All flows consist of packets with a fixed range of 500 Bytes, and the experiment lasts 70 seconds of simulated time. The offered load is asymmetric, since primarily Cluster 1 contributes 10% of the aggregate traffic, Cluster 2 contributes 20%, Cluster 3 contributes 30% and Cluster 4 contributes 40%. The complete delay and packet loss constraints for Cluster 1 are 2ms and 1% correspondingly. The comparative division issues are set of four for the ratio of delays of two successive clusters, and of four for the ratio of failure rates of two consecutive clusters. We compared the performance of buffering approach along with standard service scheduling currently in use and clustered buffering and dynamic service scheduling with priority ranks that we proposed. The same load is agreeing to all the paths with a regular interval of mille sec. Given load in kbs is shown in fig 6.2. The fig 6.3 concludes the step up of proposed QoS centric clustered buffering and dynamic service scheduling approach over buffering and service scheduling model in use. The Frame delivery ratio evaluation between proposed and existing models is compared in fig 6.4 that elevate the performance scalability as minimum Frame Overhead in the clustered 30

5 Figure 2: Weight in bytes sent by source node of the path [in regular interval of mille sec] Figure 3: Frame delivery ratio: between cluster buffering with dynamic service scheduling and single buffer strategy with standard service scheduling Figure 4: Frame Overhead comparison between proposed and existing models 31

6 5. CONCLUSION This paper proposed a Clustered conduction of VoIP routing (CCVR) topology for wireless WLAN networks. We estimate the performance of the proposed model during wide simulation study with traffic of audiovisual data combination. Our simulation study has revealed that the proposed algorithm attain considerable performance beneath QoS issues. The projected clustered buffering and adaptive flow scheduling model is exceedingly flexible in that it can offer a large range of QoS features. In future this work can be advanced such that the proposed model is qualitative in mathematically prioritizing and scheduling services. REFERENCES [1] Mohamed H. Ahmed Scheduling of Multimedia Traffic in Interface-limited Broadband Wireless Access Proc. of the 5th International Symposium in Wireless, Personal Multimedia Communications (WPMC 02), Honolulu, Hawaii, U.S.A., October [2] Mohamed H. Ahmed Scheduling of Multimedia Traffic in Interface-limited Broadband Wireless Access Proc. of the 5th International Symposium in Wireless, Personal Multimedia Communications (WPMC 02), Honolulu, Hawaii, U.S.A., October [3] William K. Wong and Victor C. M. Leung Scheduling for Integrated Services in Next Generation Packet Broadcast Networks Proc. IEEE WCNC, New Orleans, LA, Sep [4] Xin Liu Transmission Scheduling for Efficient Wireless Utilization IEEE INFOCOM [5] R. Ayala, K. Basu, and S. Ellliott, Internet Technology Based Infrastructure for Mobile Multimedia Services, Proceedings of WCNC, New Orleans, LA, September 1999,pp [6] Richardson, Sieh, Ganz Quality of Service Support for Multimedia Applications in Third Generation Mobile Networks Using Adaptive Scheduling Journal of Real Time Systems, Vol. 21, No. 3, pp , Nov