Team dionysos

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Other Grants and Activities

Section: New Results

QoS support and QoE-based ressource management in wireless networks

Participants : Kandaraj Piamrat, Adlen Ksentini, César Viho.

Supporting QoS in wireless networks is an important challenge due to the probabilistic nature of the employed MAC protocol, the well-known Carrier Sense with Multiple Access (CSMA/CA). Among the proposed solutions to sustain QoS in IEEE 802.11, we have the restriction of the accepted traffics through admission control algorithms. This protects and maintains service quality for the admitted traffic. However, if there are no restrictions to limit the traffic volume being introduced to the service set, performance degradation will result due to higher backoff time and collision rate. In [71] , we presented a delay-based admission control algorithm in IEEE 802.11. We presented an accurate delay estimation model to adjust the contention window size in real-time basis by considering key network factors, MAC queue dynamics, and application-level QoS requirements. Based on the abovementioned delay-based CW size we introduced a fully distributed admission control protocol to guarantee QoS.

Another way to guarantee QoS in IEEE 802.11 for multimedia applications is to consider cross-layer solutions that involve the application and MAC layers. In [44] , we presented a cross-layer solution to enhance VoIP in 802.11 (known also as VoWLAN). According to the MAC layer information that represent the network load as well as the wireless channel quality (in term of BER), the application layer (here the voice coder) adapts the encoding voice rate. Thus, the application layer can adapt its rate accroding to the wireless network characteristics in term of network load and physical rate.

QoS technical parameters fail to highlight user satisfaction, also called Quality of (User) Experience (QoE). Therefore, many techniques have been developed in order to assess as accurately as possible this perceptual quality. To investigate QoE measurement, we consider three approaches, namely the subjective approach, the objective one, and the hybrid method developed in the team, PSQA (see  6.10 ). In [57] , we studied loss pattern in wireless network and then measured how loss distribution has affected quality perception seen by user. We demonstrate that PSQA provides good estimations comparing to the well-known objective metric called Peak Signal to Noise Ratio (PSNR). We also observe that PSQA gives similar result compared with subjective tests (that is, with human evaluations) in most of the cases. One of the objectives of this evaluation is to validate PSQA the use of QoE as a metric for resource management in the future. For that, we give some possible directions allowing us to manage network resources using this metric. This study of QoE behavior has been conducted on a real platform in collaboration with VTT Technical Research Centre of Finland.

Recently, mobile networking has empowered our lifestyle in many ways. Wireless multicasting is one of them; it is spreading because of various applications such as mobile auction, entertainment services, etc. However, wireless multicast has some drawbacks. In order to reach farther stations and to minimize disturbance and interference, the transmission rate has been set to the lowest rate. This is a crucial point for a network operator who wants to manage valuable bandwidth and to waste as the smallest possible amount of it. We propose to adapt the transmission rate of access points (AP) according to the QoE of multicast client encountering the worst quality. If a client has a degrading QoE (less than a preferable threshold), then we lower the transmission rate. If all clients are satisfied, AP tries to increase the transmission rate in order to get more throughput. This strategy has been presented in [55] . Furthermore, this mechanism has been improved in order to adapt not only to QoE but also to varying network condition; hence the proposition of dynamic adaptation mechanism in [56] . The strategy consists in incrementing the waiting (backoff) time in binary exponential manner after each failed attempt of rate increase. This implies that the AP will have to wait longer before switching to higher rate after each failure of rate increase.


Logo Inria