3.2 HSPA network simulator design
3.2.1 HSDPA network simulator model design and development
OPNET simulation environment. The main protocols and network structure for this simulator are shown in Figure 3-1 which shows the simplified protocol architecture specifically designed for the transport network based performance analysis. However, it also provides broader coverage of performance analysis for HSDPA due to the implementation of all UTRAN and end user protocols. Therefore, in addition to the transport network layer (TNL) performance analysis, it allows Radio Link Controller (RLC), Transmission Control Protocol (TCP) and application layer protocols performance analyses and thus provides better overview about the end user performance.
Figure 3-1: simplified HSDPA network simulator overview
When the proposed network and the real system are compared, many network entities and relevant protocol functionalities are modelled in a
simplified manner since they are not the main focus of the investigation. For example, the Internet nodes and the core network protocol functionalities are simplified and added to the RNC node itself. The Radio Access Bearer (RAB) connections are provided by the three upper layers (Application, TCP/UDP and IP) of the proposed RNC protocol stack. They are representing the load generated by the core network and the Internet nodes to the transport network. The rest of the protocols in the RNC protocol stack are the transport network protocols, namely Radio Network Control (RLC), Medium Access Controller in downlink (MAC-d), Frame Protocol (FP) and the ATM based transport layers.
Figure 3-2: Detailed HSDPA network node models’ protocol stack
The user and cell classification are done in Node-B. The HSDPA simulator was designed in such a way that one Node-B can support up to 3 cells – each with a maximum of up to 20 users. In addition to the transport layers, the Node-B consists of MAC-hs and FP layers as shown in Figure 3-1. The physical layer of Node-B is not modelled in detail within the simulator. The physical layer data rate for each UE is considered in the MAC layer as a statistical approach. They are statistically analysed and emulated at MAC level as the relevant per-user data rates. With the help of the MAC-hs
scheduler, individual user data rates which emulate the wireless channel behaviour are scheduled on TTI basis. The number of MAC PDUs is selected for each user from the corresponding MAC-d probability distributions taken from dedicated radio simulation traces. Further details about the HSDPA scheduler implementation and related functionalities are given in the section 3.3.
Figure 3-2 shows the more detailed network node model of the HSDPA simulator. As depicted in the figure, the network entities and UTRAN protocols are implemented on top of the OPNET ATM node modules. ATM advanced server and workstation modules are selected from the existing OPNET modeller modules and modified in such a way that they complete the simplified HSDPA protocol architecture. The protocols RLC, MAC-d and FP are added to the ATM server node model. The key Node-B protocols FP and MAC-hs and the user equipment protocols MAC and RLC, are implemented in the ATM workstation node model. In order to analyse the effects of the transport network for the end user performance, all above newly added protocols were implemented according to the specification given by 3GPP and ATM forum [22, 62].
The overview of the detailed implementation of the MAC-d user data flow is shown in Figure 3-3. The main functionalities of RNC and Node-B protocols are shown in this figure. In summary, the user downlink data flow works as follows. The IP packets received from the radio bearer connection in the RNC entity are segmented into fixed sized RLC PDUs by the RLC protocol. Then these RLC PDUs are stored in the RLC transmission buffers until they are served to the transport network. The FP protocol handles the data flows within the transport network for each connection and this is done based on the flow control and congestion control inputs. The flow control and congestion control provide the information about the granted data rates (credits/interval) over the transport network for each UE flow independently. The FP protocol transmits the data packets from the RLC transmission buffers to the transport network based on the receipt of these flow control triggers at the RNC entity. All ATM based transport protocols transfer the data from RNC to Node-B over the Iub interface. The received data at the Node-B are stored in the MAC-hs buffers until they are served to the corresponding users in the cell by the Node-B scheduler. Once packets are received by the UE entity, they will be sent to the upper layers after taking protocol actions at the respective layers.
When the packets reach the UE protocols, the user data flow from RNC to UE is completed. For each RAB connection, the same procedure is applied from RNC to UE. The simulation model implements the data flow of the HSDPA user-plane. The control-plane protocols are not modelled within the simulator. Further details about the HSDPA simulator can be found in [3, 5].
3.2.2
HSUPA network simulator model design and development
The HSUPA network simulator is designed by adding the uplink protocols to the existing HSDPA simulator. Both simulators together form the HSPA simulator.Figure 3-4: simplified HSUPA protocol stack
Some of the protocols which are already implemented in the HSDPA simulator can be used for new uplink simulator as well. The new protocols which should be added to the existing downlink simulator for the uplink data flow are shown in Figure 3-4. It can be seen that RLC, MAC-d, MAC-es and MAC-e protocols are implemented in the UE entity, MAC-e and FP
protocols are implemented in the Node-B, and the corresponding peer to peer functionalities of RLC, MAC-d, MAC-es and FP protocols are in the RNC entity. All these protocols were also implemented in the simulator based on the 3GPP specification [22]. All the above HSUPA related protocols are implemented in a separate layer module (between IP and IPAL=IP adaptation Layer) in the workstation node and in the server node. The separate process module is named the HSUPA process module or HSUPA layer for the simulator.
The E-DCH scheduler is the main entity which models the air interface functionality in HSUPA. It is implemented in the MAC-e protocol of the Node-B. Further details about the MAC modelling are given in 3.3.So far, a brief overview about the model design and development of HSUPA and HSDPA has been discussed. Providing complete details about the implementation of all protocols in the simulator is not the focus of this dissertation. The idea is to provide a good understanding about the simulators for transport network analysis. However, some details about the key protocols which are directly related to the transport network are elaborated in the next section of this chapter.
3.3 HSDPA MAC-hs scheduler design and implementation
The exact modelling of the air interface is a very complex topic. There are several main considerations such as propagation environments with short and long term fading as well as interference which is caused by other users of the same cell and of other cells (inter-cell interference) have to be considered during this part of modelling. Further, besides the radio propagation modelling, the complete W-CDMA air interface with scrambling and spreading codes, power control etc. needs also to be modelled within the physical layer. Modelling of such detailed radio interface is a part of the link level simulator. From the system level simulator point of view, modelling the complete air interface is not practicable, so it only uses an abstract model of the air interface with certain accuracy of the system level performance. Therefore, the complete WCDMA radio interface for HSDPA is modelled in combination with the MAC-hs scheduler modelling.