On the Design of Robust and Adaptive IEEE Multicast Services for Video Transmissions Speaker: Bo-Yu Huang Advisor: Dr. Ho-Ting Wu Date: 2014/12/23

Outline  Introduction  The current trends on multicast video transmission in IEEE WLANs.  Adaptive Multicast Mechanism with Collision Prevention (AMM/CP)  Simulation results  Conclusion  References

Introduction  Using wireless links for video streaming has become more common. However, real-time video streaming over wireless networks is a challenging proposition due to the characteristics of the video data and wireless channels.  In wireless environments, the channel conditions change rapidly over time due to noise, interference, multipath, and user mobility. In such context, the transmission control schemes have to dynamically adapt the transmission rate to the channel conditions.

Introduction(Cont.)  In the current IEEE standard, there are two main inter-related issues that need to be addressed on the design of a reliable and scalable multicast service: Provisioning of an efficient and scalable feedback mechanism: Multi-rate transmissions.

Introduction(Cont.)  In order to enhance the performance of the multicast video service provided by the IEEE WLANs, we propose a novel mechanism, referred from now on as Adaptive Multicast Mechanism with Collision Prevention (AMM/CP).

Current trends on multicast in IEEE  Most research efforts on multicasting in IEEE WLANs have focused on improving the reliability of the multicast service by integrating Automatic Repeat-reQuest (ARQ) mechanisms into the protocol architecture. Leader-Based Protocol(LBP)

 This protocol assumes that one of the receivers of the multicast has been chosen to be a leader for the purpose of supplying CTS and ACK in response to RTS and data packets.

 [A] Base -> Receivers Send multicast-RTS.  [B] Receivers -> Base Leader: if ready to receive data, send CTS. if not ready to receive data do nothing. Others: if ready to receive data, do nothing. if not ready to receive data, send NCTS (Not Clear to Send).

 [C] Base -> Receivers If a CTS was heard in slot 2, start multicast transmission. If no CTS was heard in slot 2, back off and go to Step A.  The next step is executed only when multicast transmission occurs in Step C.  [D] Receivers -> Base Leader: if packet received without error, send ACK. if in error, send NAK. Others: if packet received without error, do nothing. if in error, send NAK.

 LBP uses both ACKs and NAKs from receivers as feedback to the sender. It allows collision of an ACK with one or more NAKs to ensure that the sender does not get a positive feedback if one or more group members receive erroneous transmission.

 Even though the LBP address the reliability problems, it falls short on providing a valuable solution to the adaptive channel rate adaptation based on the changing channel conditions. To solve the rate adaption problem, theses schemes can be used together with the ARF protocol.  The Probing-based ARF (PARF) mechanism. PARF attempts to provide a suitable solution to the adaptive transmission rate issue, however its performance statistics do not defer substantially from the ones reported for the LBP+ARF. The main drawback is that they do not provide a complete solution to overcome the main error sources, i.e., channel access conflicts and channel varying operating conditions.

 The IEEE Task Group aa(TGaa) has defined a new multicast service, named Group Addressed Transmission Service (GATS). Under this new service, each group-addressed stream may be delivered using different schemes according to the requirements of the stations belonging to each stream. This differentiation is done by allowing a station to request higher reliability for one or more group addressed streams. GATS comprises two services: Directed Multicast Service (DMS) and GroupCast with Retries (GCR).

Directed Multicast Service (DMS)  Each multicast frame is transmitted in a unicast mode to each multicast groupcast member. Frames transmitted to multicast addresses are individually transmitted to each of the associated STAs belonging to the multicast group. Those frames will be retransmitted until receiving an acknowledgement from the AP, and they will be stopped when the retransmission limit is achieved. Although this mechanism guarantees a reliable transport to multicast traffic like unicast transmissions, it has large scalability when using a high number of the multicast group

GroupCast with Retries (GCR)  GCR is a flexible service whose main aim is to improve the delivery of group-addressed packets. Unsolicited-Retry: the AP retransmits a packet one or several times to increase the probability that all the MRs in the multicast group successfully receive the packet. Block-ACK mechanism: This method extends the Block Ack mechanism specified in IEEE n for use in multicast transmissions to a group. The AP transmits a number of multicast frames and then requests from one or more of the recipients to acknowledge the receipt of the transmitted frames. Frames that have not been received correctly by one or more of the receivers can then be retransmitted.

 With the definition of the GATS, the IEEE aa offers different solutions to the reliability problem. However, this amendment does not introduce any improvement for the flow adaptation problem.  In short, the IEEE aa amendment does not specify an efficient mechanism including a rate adaptation mechanism into the multicast service.

The AMM/CP mechanism  In order to enhance the performance of the multicast service provided by the IEEE WLANs, we present a multicast mechanism for multi-rate transmissions, called AMM/CP.  The design of AMM/CP addresses the following three issues: Collision prevention Negative feedback for multicast packets Flow adaptation

Collision prevention  It integrates the MCP mechanism, having proved effective on reducing various orders of magnitude the multicast collision rate with respect to the one reported by the standard multicast mechanism

Negative feedback for multicast packets  To avoid the potential implosion of ACKs to be sent by the MRs, the protocol makes use of NAK packets instead ACKs. Thus, an MR will only issue a NAK in case of receiving a corrupted multicast packet.

Flow adaptation  AMM/CP also makes use of a modified ARF mechanism. In this way, the data transmission rate may be changed to the one offering better guarantees given the channel conditions.

The AMM/CP mechanism(Cont.)  In the AMM/CP mechanism, when the AP has a multicast packet to send, it first invokes the MCP mechanism to reduce the collision probability. Once the multicast packet is delivered, the AP senses the channel. If the AP senses the channel busy before the timeout timer expires, it assumes that one or more MRs have issued a NACK. In such situation, the AP should run the rate-adaptation mechanism to check if a more robust transmission rate should be used to retransmit the packet. On the other hand, if the waiting period expires without having detected any activity in the channel, the multicast packet has been successfully delivered to all MRs. In this case, the AP runs the rate-adaptation algorithm to decide if it increases the transmission rate of the next multicast packet.

Simulation results  A comparison of AMM/CP and a reliable multicast service incorporating the LBP and the ARF mechanisms( LBP+ARF mechanism ).  The evaluation is done both in terms of QoS guarantees and QoE offered to the end user.  QoE is evaluated using the VQM_VFD metric and the results are converted to the Mean Opinion Score (MOS) scale, whose values spans from 1 to 5 (1-bad, 2-poor, 3-fair, 4-good, 5-excellent).

Simulation results(Cont.)

Conclusion  In this paper, we have evaluated our new multicast mechanism. It is able to adapt the transmission rate of the multicast flow to the channel conditions. The combination of collision prevention, feedback and rate adaptation makes that our mechanism properly delivers the multicast video flow to a group of mobile stations. We have proved it by developing a video quality experiment for evaluating the QoS of the WLAN and the QoE perceived by the mobile members of the multicast group.

References  M. Santos, J. Villalon, L. Orozco-Barbosa, and L. Janowski, "On the design of robust and adaptive IEEE multicast services for video transmissions," in Proc. of WoWMoM, 2013, pp  J. Kuri and S. K. Kasera, “Reliable Multicast in Multi-access Wireless LANs”. ACM Wireless Networks”, Vol. 7(4), pp , 2001.

Thanks for listening