Research On Voice Quality In VoIP Based On Differentiated Services

VoIP (Voice over IP),i.e. IP phone,which is an advanced voice communication transmission technology through IP networks. In recent years, IP phone is widely used because of its great potential,and is considered as one of the goals of the next generation network (NGN). However,designed only to handle non-real-time data traffic, IP network can only provide "best effort" service which can not satisfy the Quality of Service requirement for carrying vioce and video. How to improve the real-time performance to ensure the voice communication quality is a critical technique issue and one fo the technique difficulties of IP phone system.In order to provice the assurance of Quality of Service (QoS) of the IP network, IETF proposes the DiffServ (Differentiated Service) architecture, the DiffServ have been standardized in the PHB are: best effort BE, expedited forwarding EF, assured forwarding AF and class selector CS. EF PHB is to support low loss, low latency, low jitter, assured bandwidth, end-to-end service through DS domains, so very suitalbe for VoIP to provide voice quality assurance. The thesis studies mainly studies the performance of VoIP based on DiffServ.The thesis first analysises the prospects of VoIP technique, overviews the basic principles of VoIP, analysises the main elements affecting the voice quality of VoIP, researches the methods for evaluating the quality of voicefor VoIP network, and discusses the basic technique for assuring the quality of voice for VoIP on the application level and the network level. After that analysises the principle of DiffServ. Compares the performance of VoIP both on DiffServ EF PHB and on Best-effort by simulation, through E Model, gives the quantitative assessment of the voice quality of VoIP, and carries out the user-level performance analysis for the VoIP on DiffServ EF PHB, the experimental results show that DiffServ provides better assurance than Best-Effort when network traffic congests and the set-aside resources for VoIP are adequate. However, the voice qulity of VoIP both on DiffServ and Best-Effort declines and can not achieve a user acceptable level when network traffic congests and the set-aside resources for VoIP are inadequate. To the question of DiffServ, proposes a new packet marking algorithm and realizes the algorithm in NS-2.Compares the new marking algorithm and the old marking algorithm by simulation, the experimental results show that DiffServ networks applied the new marking algorithm can assure a part of VoIP calls keeping good voice quality and achieving user satisfaction level when network traffic congests and the set-aside resources for VoIP are inadequate.