LTE EPC ARCHITECTURE

As illustrated in Figure, the EPC is made up of a control plane node called MME (Mobility Management Entity) and two user plane nodes called S-GW (Serving GW) and P-GW (Packet Data Network GW). In the non-roaming case the S-GW and P-GW functionalities are both located within one operators network and can be implemented in a combined S- and P-GW node.


The E-UTRAN is made up of eNB nodes, which are connected to each other via the X2 interface. Both the S1 and the X2 interface can be divided into control plane (dashed lines) and user plane (solid lines) parts. The S1 and X2 are multi-to-multi capable interfaces meaning that one eNB can be connected to multiple MMEs, S-GWs and other eNBs. A given terminal however can only be connected to one eNB, MME and S-GW at a given moment. Two terminals
connected to the same eNBs can be connected to different S-GWs and vice versa.


In a typical implementation of EPC the gateway nodes (P-GW and
S-GW, which together also is referred to as SAE-GW) is by including both of them in the GGSN software. The GGSN hardware may thus be used for both GGSN and SAE-GW functionality.



The MME may in a similar fashion be implemented in the SGSN software. The SGSN hardware may thus be used for both SGSN and MME functionality.



When an E-UTRAN system is deployed in a network already supporting GERAN and/or UTRAN it will be possible to use a common core network for all accesses. In practice this means that the P-GW will provide GGSN functionality towards the existing General Packet Radio Service (GPRS) CN. A EUTRAN/UTRAN/GERAN capable terminal will therefore not need to change the GGSN (i.e., the IP point of presence towards external networks) when it changes Radio Access Technology (RAT) between GERAN, UTRAN and E-UTRAN

.the UTRAN when utilizing the “GPRS One Tunnel Approach” currently being standardized in 3GPP Rel-7. This feature makes it possible to bypass the SGSN in the user plane.

This Figure shows a standardization view on how GERAN, UTRAN and E-UTRAN are integrated into the SAE CN. It should however be noted that the SGSN and MME shares a lot of common functionality. It is also required that the CN protocols, Session Management (SM) and Mobility Management (MM), used in
2G/3G are compatible with the respective protocols used in EPS meaning that the SGSN and MME share a common evolution in the 3GPP standard. In a typical implementation/deployment view, it is likely that the 2G/3G SGSN and the MME are merged into one node (as illustrated in Figure 2-12). This will make it possible to support intra SGSN/MME and inter P/S-GW/GGSN node mobility between the different accesses.