 We are going to look at the quality of service aspects of the system architecture evolution how the 3GPP guidelines for release 8 onwards concern the quality of service. We shall start with the 3GQS guidelines. Then we are going to go deeper into the advanced LTE requirements. Of course, this is all implemented through prioritizing the traffic. Let's look at what was the concept of quality of service in terms of 3G. That is UTRAN. We are talking about HSPA. We are talking about the 3GPP LTE networks. Of course, back in those days, I'm talking about like 7, 8 years or 10 years ago. We already had a heterogeneous service environment. Multiple services would compete against each other to satisfy the customer requirements. Therefore, 3GPP Alliance already had defined quality of service classes for mobile traffic. That included the UMTS and HSPA traffic. Please note that the issue of quality of service is only relevant when we are talking about data as well as voice circuits because if it is only voice then there's no prioritization. All the voice circuits have to be given the same treatment. But once we are comparing or competing voice with data, then we have to consider what is going to be the quality of service mechanism. So in 3GPP, the voice was taking a different path. That is circuit switch network bearers and then the internet traffic, which was best effort, was based on packet switching. So it means in 3GPP, we were already implementing separate connections for voice as well as for data. So as such, the quality of service requirement was only confined to the data services. This changed with the LTEA, that is system architecture evolution, because now all the traffic is based on IP. So it means now the voice is also going to be based on IP. Quality of service suddenly became more critical and harder to implement. Now we have voice based on the IP protocol stack. It means it is competing directly with data services, interactive services, and in addition to pure voice, we have voice over IP, we have IP TV, and we have gaming. So in the presence of multiple services, the quality of service provisioning becomes a greater challenge. For that, the QoS classes, which were described earlier in 3GPP up till the LTE, the quality of service class identifier was a metric, QCI, that would differentiate between different classes. It basically implements certain QoS parameters that start from traffic priority, that is, if we have multiple packets belonging to different user applications, including voice and video, which takes precedence over the other in terms of enqueuing and scheduling. Then we have the loss probability as another parameter. It says if we have multiple packets, which are at the tail of a queue and we have to drop one or some of these packets, which packets would be prevented from dropping, that is, they have a low loss probability. Packets which do not have high quality of service requirements would have higher loss probability. Then we have the delay. Delay means the end-to-end delay, which is to be implemented to meet certain real-time interactive application requirements, such as voice as such, video, interactive games, and streaming, audio and video. The resource types, that is, the services, which are defined for SAE include the guaranteed bit rate and the non-guaranteed bit rate. When we say guaranteed bit rate, it means the traffic which would be entitled to the guaranteed bit rate would be provided some minimum bit rate, at least under all conditions, whereas for non-guaranteed bit rate services, no such commitment shall be made. So depending upon their various combinations, including priority, loss probability, delay, et cetera, overall, nine QCI values have been defined. Of course, with them come nine different priority levels. The highest priority level is actually dedicated to the signaling, because signaling serves as the core functionality without which services cannot be provisioned. The IP multimedia sub-system, which is the heart of the signaling system in the overall next generation network, 3GPP architecture. The priority is given to the IMS signaling through the non-guaranteed bit rate resource type. It means it is given a higher priority, but no prior commitment is made to a specific data rate requirement. And it goes without saying that if there is no signaling, no connections shall be established and no data can be serviced. Once we have given the highest priority to signaling, then we have the priority levels going from the high to low. First of all, we have the guaranteed bit rate services, which are prescribed for voice over IP, audio, video, calling, streaming, gaming. And for non-guaranteed bit rate, we have the typical TCP IP based applications, which are also known as elastic applications. These could include the interactive games, which typically people play online, the browsing, email, file transfer, etc.