DHCP-based Enhanced Wireless Access Gateway Overview

This chapter describes the DHCP-based Enhanced Wireless Access Gateway (D-eWAG) solution.

The following topics are covered in this chapter:

Product Overview

The D-eWAG solution described in this chapter is designed for centralized WLAN deployments, wherein Access Points (APs) spread across geographical locations provide Wi-Fi access, and Wireless LAN Controllers (WLCs) located in a central server farm control all the APs.


Figure 1. D-eWAG Deployment

The D-eWAG acts as first-hop L3 router to WLC with direct connectivity between them and is located in the central server farm. With the use of Service Set Identification (SSID)-based WLAN access, subscribers can be authenticated based on the SSID that they use in order to connect to the WLAN. The AP/WLC maintains a separate SSID for providing 3G access. This enables the UE to select the correct SSID for obtaining 3G access through the Wi-Fi network.

The D-eWAG also acts as the AAA Proxy and the DHCP server to the UE attaching to the WLAN network. This helps in processing all the control packets from the UE and maintaining the subscriber session to provide 3G access. While acting as DHCP server, D-eWAG creates the PDP-Context with GGSN to obtain the IP address to be allocated to the UE through DHCP-Response in the access side. Note that this interface with GGSN is similar to the TTG's Gn' interface with GGSN in 3GPP.

When the UE wants to gain 3G access through the Wi-Fi network, the subscriber selects the 3G-SSID from the list of advertised SSIDs.

The WLAN attach procedure occurs in three stages:

  1. Association process
  2. 802.1x EAP-SIM/AKA authentication process
  3. IP address allocation process

These three steps are transparent to the subscriber accessing the Wi-Fi network and do not involve any subscriber intervention. At the end of the WLAN attach procedure, the UE connects to the 3G network.

Deployment Models

The D-eWAG can be deployed in any of the following ways:

  • Stand-alone D-eWAG deployment on an ASR 5000 chassis.
  • Combo D-eWAG + GGSN deployment on the same ASR 5000 chassis.

IMPORTANT:

In this release, the following deployment option is not qualified and is not supported, it is available only for lab testing purposes.

  • Combo D-eWAG + R-eWAG deployment on the same ASR 5000 chassis.

IMPORTANT:

For assumptions and dependencies pertaining to the network models discussed in this section, refer to the Dependencies and Limitations section.

Supported network deployment models:

  • One SSID mapped to one VLAN mapped to one APN.
    • Each SSID should always be mapped to a unique VLAN in this case, even if it is served using multiple WLCs.
    • Different VLAN used for all UE sessions connecting through different SSIDs and uplink packets can be identified uniquely with {VLAN+Source IP} at D-eWAG.
  • One SSID mapped to one VLAN mapped to multiple APN.
    • Each SSID should always be mapped to a unique VLAN in this case, even if it is served using multiple WLCs.
    • Same VLAN used for all UE sessions and so the uplink packets cannot be identified uniquely with {VLAN+Source IP} at D-eWAG as there can be overlapping IP addresses in this case.
    • This type of deployment is needed to ensure that the multiple APNs being served do not contain overlapping IP address space.
  • One SSID mapped to multiple VLAN mapped to one APN.
    • WLCs can be different with different VLAN for same SSID.
    • WLC can be configured with AP-Group to use different VLAN.
    • Set of VLANs serving one APN are different from set of VLANs serving another APN. Hence, overlapping IP address is not an issue in this case as the session can be identified uniquely using {VLAN+Source IP}.

3G-SSID

The SSID created in Wi-Fi network for 3G access through D-eWAG is referred to as 3G-SSID. The following options (not restricted to) can be considered for 3G-SSID creation in Wi-Fi networks:

  • Each SSID (or WLAN) represents particular APN network access of an operator. One SSID per APN case.
  • Each SSID (or WLAN) represents particular operator itself. This is one SSID per operator scenario where multiple APN served by that operator can be accessed through this SSID. This means that the different users connecting through this SSID can be subscribed to different APN served by that operator. All the users can gain access to their subscribed APN network as the 3GPP-AAA server will return the subscribed APN to D-eWAG and selects GGSN based on that.

Association Process

During the 802.11 Association process, the access points allocate resources for UE communication and synchronize with the UE. This is as per the standard 802.11 process and D-eWAG is not involved in this process.

802.1x EAP-SIM/AKA Authentication Process

After the association process has completed:

  • AP/WLC asks for UE identity by sending EAP-ID request through 802.1x authentication. Both EAP-AKA and EAP-SIM authentication methods are supported in this model.
  • UE sends its EAP-Identity in the form “IMSI@realm” in EAP-ID-Response message.
  • This EAP-ID-Response message is sent to the AP/WLC where it creates the corresponding RADIUS Access-Request to the AAA Server. Note that the AAA server for this 3G-SSID is D-eWAG. Thus, the Access-Request message is sent to D-eWAG over the VLAN mapped to that 3G-SSID (3G-WLAN) from WLC.
  • D-eWAG acting as AAA-Proxy uses this RADIUS Access-Request message and uses the same as First Sign of Life (FSoL) for UE session creation and stores the UE's MAC address (Calling-Station-ID) to uniquely identify the session.
  • D-eWAG selects the 3GPP-AAA server for UE authentication based on the realm part received in the user-identity (inside RADIUS Access-Request) and proxies the Access-Request to that server. If the realm part is not available in the EAP-Identity, then the locally configured default 3GPP-AAA server is selected. This way the normal EAP-SIM/AKA authentication procedure will continue between UE and 3GPP-AAA server with D-eWAG acting as AAA-Proxy.
  • At the end of the authentication procedure, D-eWAG caches all the 3GPP-specific parameters used for PDP-Context Creation with GGSN (like MSISDN, APN, Charging-Char, etc.) from the Access-Accept message. The 3GPP-AAA server sends all the 3G attributes in the Access-Accept message (similar to PDG/TTG in 3GPP).

IP Address Allocation Process

After successful authentication using 802.1x in WLAN, the UE initiates the DHCP signaling message to obtain the IP address. The WLC should be configured as DHCP-Relay-Agent and the D-eWAG IP address should be configured as the external DHCP-Server at WLC for 3G-SSID. The DHCP-Discover broadcast message from UE is processed by WLC (DHCP relay) and sent as Unicast DHCP-Discover Request to D-eWAG (DHCP-Server) over the mapped VLAN. This DHCP-Discover message contains the CHADDR field containing the UE's MAC address and helps in identifying the correct session uniquely at D-eWAG. After the UE session is identified, D-eWAG initiates the PDP Context Creation procedure with GGSN and obtain the IP address. Note that the 3G attribute used for the creation of PDP-Context was already cached at D-eWAG during the authentication process. D-eWAG sends the DHCP-Offer message with the IP address allocated by the GGSN set in the “Your-IP-Address” field. The subsequent DHCP-Request message from the UE containing the GGSN-allocated IP address is acknowledged with the DHCP-Ack message by D-eWAG. This way the UE gets the WLAN IP address directly from the 3G network and starts sending data traffic.

The following additional host configuration parameters should be provisioned for the UE during DHCP signaling since the access is WLAN:

  • Default gateway
  • Subnet mask/prefix length
  • DNS server address
  • DHCP server address

After the WLAN attach procedure is completed as explained above, D-eWAG session for the UE becomes active and ready for data transfer.

Note that if the WLC sends the Accounting-Start message to the D-eWAG (if it is configured as Accounting-Proxy at WLC), it will proxy the Accounting-Start message to the 3GPP-AAA server and send the Accounting-Response message back to the WLC.

Data Traffic between WLAN and 3G Network

As the D-eWAG acts as default-gateway for the UE, all uplink data packets are received by D-eWAG and sent to the GGSN over GTP-U tunnel. When the downlink data packet is received from GGSN over the GTP-U tunnel, D-eWAG throws the packet to WLC over the VLAN mapped for the UE session, and WLC delivers the packet to the UE.

D-eWAG as First-Hop Router to WLAN Network

The D-eWAG acts as the first-hop router to the WLAN network, which provides access to the 3G domain. This means that D-eWAG has L2 connectivity with the Wireless LAN Controller (WLC) using VLANs and acts as first-hop router to route traffic to the GGSN.

In a typical Wi-Fi network each SSID will have corresponding VLAN mapping at WLC node. Therefore, the network setup should be in such a way that D-eWAG should also be the member of all VLANs as that of the WLC's VLAN serving 3G-SSID. This ensures that all the traffic from UEs attaching to any 3G-SSID will reach the D-eWAG acting as first-hop router through WLC.

Each VLAN interface at D-eWAG can be connected to one or more WLCs serving the same SSID, and each WLC acts as RADIUS client and DHCP relay for that SSID. So, the RADIUS-client/DHCP-relay function at WLC will use the IP address of VLAN interface mapped to that 3G-SSID, and D-eWAG is configured as corresponding RADIUS/DHCP server.

D-eWAG as Default Gateway

D-eWAG operates as first-hop L3 router (default-gateway) for Wi-Fi clients (UE), it should be possible for all UEs to send data traffic directly to the D-eWAG. This is achieved by sending the default-gateway DHCP option (or DHCP ROUTER option-3) as described in the Requirements at GGSN section. Note that this default-gateway IP address should be in the same subnet as that of IP address allocated by the GGSN.

Thus, when the UE wants to send traffic, it will first resolve the MAC address of the default-gateway using ARP-Request. This ARP-Request gets forwarded by WLC over mapped VLAN and D-eWAG responds with ARP-RESPONSE as it owns the IP address. This ensures that all the data packets from the UE reach D-eWAG.

When the default-gateway configuration is not available or does not match with subnet of the allocated IP address from GGSN, the call will get dropped. This ensures that any consecutive DHCP packets from that UE get dropped at D-eWAG.

APN Selection

APN for the D-eWAG session is selected in following way:

  • APN for a particular session is returned by the 3GPP-AAA server during authentication. The APN can be sent using the RADIUS “Service-Selection” AVP in Access-Accept message from the 3GPP-AAA server.
  • If the APN is not supplied during authentication, the locally configured APN under the subscriber-template configuration is applied to the D-eWAG session.

D-eWAG Service in the ASR5000 Chassis

D-eWAG’s service capabilities include:

  • The D-eWAG service acts as an authentication-proxy during authentication of UE with 3GPP AAA. This is to process authentication messages between the UE and 3GPP-AAA server and to obtain the 3G-specific attributes required for PDP context creation with the GGSN.
  • D-eWAG service acts as DHCP server terminating the DHCP-Relay messages from the AP/WLC. This is to process the actual DHCP signaling during Wi-Fi attach procedure and return the IP address allocated by GGSN (during PDP context creation) in DHCP message itself.

    IMPORTANT:

    Note that the DHCP service must be configured in DHCP-Server mode in the same context as the D-eWAG service.

  • D-eWAG acts as accounting-proxy to proxy the RADIUS accounting messages between WLC and 3GPP-AAA.

WLC - D-eWAG Interface

As discussed earlier, the interface between WLC and D-eWAG is based on VLAN. Note that there can be multiple WLCs connecting to a single D-eWAG. In which case, each WLC should be part of at least one VLAN which is shared by D-eWAG. This helps the control/data packets from 3G-SSID reach D-eWAG from WLC through that VLAN.

Control Plane

Following are the control signaling packets to be handled by D-eWAG during the WLAN attach procedure by UE in the 3G-SSID WLAN network:

  • 802.1x authentication
  • DHCP IP assignment
  • RADIUS accounting

Requirements for 802.1x Authentication

  • Ingress EAP authentication messages are all encapsulated inside RADIUS messages.
  • WLC configured with D-eWAG service IP address as the AAA authentication server for the 3G-SSID.

Characteristics of this control flow:

  • D-eWAG acts as AAA-Proxy for the authentication happening between UE and 3GPP-AAA.
  • D-eWAG selects the actual 3GPP-AAA server based on REALM part in the NAI received in “Username” AVP. This is achieved using the Subscriber Template based operation of D-eWAG in the ASR5000 chassis.
  • The first inbound RADIUS message (Access-Request) is the FSoL for D-eWAG to create a new D-eWAG session.
  • The UE MAC address present in the “Calling-Station-ID” AVP of Access-Request message is used to identify the UE session at D-eWAG for subsequent RADIUS messages from the WLC.
  • At the end of 802.1X authentication, the Access-Accept message from 3GPP-AAA server carries the 3G-specific attributes of the authenticated user such as IMSI, MSISDN, and APN. This information is used by D-eWAG for creating a GTP PDP context with the GGSN.

DHCP Requirements

  • The WLC should act as DHCP-Relay and should be configured with D-eWAG service IP address as the “external dhcp-server” for the 3G-SSIDs.
  • D-eWAG processes all the DHCP messages sent to standard DHCP server UDP port 67.
  • When DHCP-Discover message is received from the UE, DHCP server in the ASR5000 chassis goes into pending state to wait until the signaling on the MNO side (GTP tunnel creation) is done to get an IP address for the UE.
  • On the arrival of the Create PDP Context Response, which carries the assigned IP address c.c.c.c for the client, DHCP is fully resumed to offer c.c.c.c back to the client.
  • On the completion of DHCP signaling, the session on the DP is fully activated to tunnel the client's entire traffic to the GGSN over GTP-U.
  • In subsequent DHCP message exchanges over time (for example, DHCP Request and DHCP ACK), no further signaling will happen on the MNO side. The DHCP-REQUEST on the D-eWAG needs to always turn around to compose a corresponding response to reassign or renew this same address with an endless lease back to the client.
  • IMPORTANT:

    UE suggesting the IP address to DHCP server in DHCP-Discover or DHCP-Request messages is not supported in this release.

  • UE connecting through D-eWAG should include the “PARAMETER REQUEST LIST” DHCP option in DHCP-Discover/Request to ask for subnet-mask, default-router, and DNS configuration parameters from DHCP Server (D-eWAG) as DHCP-Inform message is not supported in this release.
  • DHCP service should be configured in the same context as the D-eWAG service. This is because D-eWAG is using the existing DHCP service in the ASR5000 chassis to act as DHCP-server in this model.

RADIUS Accounting

RADIUS accounting messages are exchanged in the WLC-D-eWAG interface as described here:

  • WLC node can be configured with D-eWAG service IP address as the RADIUS accounting-server for the 3G-SSID sessions.
  • After the IP address is allocated to the WLAN UE using DHCP signaling, WLC will send the RADIUS Accounting-Start/Interim/Stop messages for the UE session to D-eWAG.
  • The accounting messages received are proxied to the 3GPP-AAA server (like authentication process) by D-eWAG. Acct-Interim message are used for D-eWAG session updates like identifying AP change, and Acct-Stop message are used to teardown the D-eWAG session as the corresponding session at WLC is down.
  • Note that this accounting proxy is optional. WLC can have different AAA server configured for RADIUS accounting.
  • When D-eWAG receives a RADIUS accounting message from WLC, it is forwarded to the AAA server. In this scenario, if the call goes down for any reason apart from Acct-Stop from WLC, D-eWAG creates Acct-Stop on its own for this WLC-initiated accounting and sends it to the AAA server. This ensures that the AAA server will know that the WLC-initiated accounting session needs to be stopped as the session has gone down. However, if there is no accounting message received for that session from WLC then D-eWAG will not send Acct-Stop on its own for WLC accounting session on call teardown.

D-eWAG - AAA Interface

By acting as AAA Proxy, D-eWAG will be proxying all the RADIUS authentication/accounting messages between AP/WLC and the 3GPP AAA server. D-eWAG selects the actual 3GPP-AAA server based on REALM part in the NAI received in “Username” AVP. D-eWAG operates based on the Subscriber Template in ASR5000 chassis and thus the AAA server is selected.

RADIUS CoA/DM Support

RADIUS CoA

D-eWAG supports CoA messages from the AAA server to change data filters associated with a subscriber session as well as QoS value, rulebase, and Firewall-NAT-policy.

The CoA request message from the AAA server must contain attributes to identify NAS and the subscriber session and either filter rule, Firewall-NAT-policy or QoS or rulebase name.

If the system successfully executes a CoA request, a CoA-ACK message is sent back to the RADIUS server and the data filter is applied to the subscriber session. Otherwise, a CoA-NAK message is sent with an error-cause attribute without making any changes to the subscriber session.

IMPORTANT:

Note that D-eWAG does not forward the CoA request to WLC. WLC does not support CoA.

IMPORTANT:

Changing ACL/rulebase/Firewall-NAT-policy/QoS together in a single CoA is not supported. For this, separate CoA requests can be sent through the AAA server requesting for one attribute change per request.

Filter-ID

The “Filter ID” AVP contains name of the data filter to apply to the subscriber session. The “filter-id” attribute (attribute ID 11) contains the name of an Access Control List (ACL).

QoS

If CoA is received with QoS value the same is sent to GGSN in UPC Request and on receiving successful UPC Response, CoA Ack is sent. Otherwise, CoA-Nack is sent.

Firewall Policy

CoA if received with Firewall policy name must be applied to the subscriber session. If the system does not support that Firewall policy for the subscriber then CoA-NACK is sent.

Rulebase

CoA can have Rulebase AVP to specify new rulebase to apply to subscriber.

RADIUS Disconnect Message

RADIUS Disconnect Message (DM) is used to disconnect subscriber session in the system from a RADIUS server. The DM Request message contains necessary attributes to identify the subscriber session.

If the system successfully disconnects the subscriber session, a DM-ACK message is sent back to the RADIUS server, otherwise a DM-NAK message is sent with proper error reasons. If disconnect ACK is sent then as per normal deallocation path D-eWAG sends disconnect request to WLC as well (if configured in the D-eWAG service configuration).

IMPORTANT:

Disconnect Request sent by the D-eWAG to the WLC may not contain the same attribute list that it received in Disconnect Request from 3GPP AAA.

RADIUS Accounting Support

D-eWAG supports RADIUS accounting. It uses subscriber template configuration to obtain accounting mode information.

D-eWAG - GGSN (Gn')

The Gn' reference point is between the D-eWAG and the GGSN. Here the D-eWAG acts as an SGSN and initiates the creation of a PDP context.

For every UE, the D-eWAG creates one GTP tunnel with the GGSN. The W-APN, IMSI, MSISDN, Charging Characteristics, and QoS of the WLAN-UE are forwarded to GGSN in Create-PDP-Context-Request message.

GGSN Selection

The GGSN node is selected as per the 3GPP standard of resolving the IP address using DNS query.

This DNS query contains the DNS-APN string in the form “<apn-name>.mncXXX.mccYYY.gprs”. The APN name is derived from either local-configuration or obtained from AAA server during Access-Accept message. MCC and MNC values are derived in the following priority:

  1. From the NAI sent by UE in Access-Request message in the form “IMSI@wlan.mncXXX.mccYYY.3gppnetwork.org”.
  2. Local configuration.
    Configured using the plmn id mcc mcc mnc mnc CLI command under the D-eWAG service.

GTP Messages

The following messages are supported over the Gn' reference point:

  • Create PDP Context Request/Response.
  • Update PDP Context Request/Response:
    • GGSN-initiated UPC handled for updating QoS.
      GGSN-initiated UPC Request is accepted only for QoS Update case. QoS is updated for the D-eWAG session and accept status is sent in UPC Response. UPC Requests with EUA Update, PCO Update, APN Restriction Update, TFT Update, Direct Tunnel Update will be rejected by D-eWAG.
      Note that only EUA Update rejection from D-eWAG will cause session teardown at GGSN and subsequently D-eWAG session will be torn down through GGSN-initiated DPC. Also, note that EUA Update is sent by GGSN in UPC Request only when GGSN had sent 0 IP address in EUA IE of the CPC Response.
    • D-eWAG-initiated UPC when new AP Location Information is received in the Accounting-interim message for the session, and when COA with QoS update is received from 3GPP AAA.
      UPC response handling scenarios:
      • If GGSN responds with UPC failure with cause other than “non-existent”, there will be no QoS update for the D-eWAG session. The session persists in this case.
      • If GGSN responds with UPC failure with cause set to “non-existent”, the D-eWAG session gets removed. Disconnect Message is sent to the WLC.
      • If there is no UPC response from GGSN, GTP path failure is assumed and the D-eWAG session is removed.
  • Delete PDP Context Request/Response
  • Error Indication
  • Version Not Supported
  • GTP Payload Forwarding
  • GTP Echo

IMPORTANT:

As the WLC cannot send 3gpp-qos, UPC from D-eWAG to GGSN for QoS change from WLC does not happen.

IP Address Allocation

Dynamic IP Address Allocation

In this case, IP address for the UE connecting through WLAN is dynamically assigned by the GGSN. As explained earlier, the UE initiates DHCP-Discover to obtain IP address after authentication. D-eWAG creates the PDP-Context in response to this DHCP message. The End-user-address IE in the Create PDP Context Request message (indicating dynamic address assignment by GGSN) is empty, which makes the GGSN allocate an IP address in the Response message.

Static IP Allocation

IMPORTANT:

Static IP Allocation is not supported in this release. D-eWAG responds to DHCP static IP request with DHCP NAK.

The UE can sometimes request for an IP address using the “requested ip address” (option 50) field in DHCP message. The scenario could be that the UE was earlier attached to the 3G network using macro-cell and is now connecting through WLAN. Thus, it will try to retain the IP address it was allocated during 3G access by requesting the same through DHPC message. In this case, D-eWAG will also request for the same IP address to GGSN by filling it in the “End-user-address” IE in CPC Request. If the GGSN is not able to allocate the requested IP address, then D-eWAG drops the call and DHCP-Offer message is not sent back.

NSAPI Allocation

D-eWAG is responsible for allocating NSAPI values before sending the Create-PDP-Context-Request message to the GGSN. Although the D-eWAG acts as an SGSN in terms of GTP tunnel establishment, it also manages NSAPI allocation as WLAN UEs do not send NSAPI in this case. The default NSAPI allocated by D-eWAG is 15.

UE Identity and Location Information Support

The D-eWAG supports sending UE identity and location information to the GGSN, which the GGSN can use for Lawful Intercept support.

RAI

The RAI IE in CPC Request sent to GGSN is encoded using the MCC MNC or PLMN ID configured at D-eWAG.

ULI

The User Location Information (ULI) IE in CPC Request sent to GGSN is encoded using the “Called-Station-ID” AVP received in Authentication-Request message at D-eWAG. The “Called-Station-ID” AVP contains the Access Point Identifier (AP Identifier), which is composed of the Location Area Code Cell Identity (LAC_CI) — that is, Location Area Code (LAC) and Cell Id (CI) separated by an underscore. For example, if the access point is assigned LAC = 123 and CI = 56789, then the “Called-Station-ID” AVP will contain 123_56789. As per 3GPP TS 23.003, the LAC and CI are each 2 bytes in length.

Note that the “Called-Station-ID” AVP is optional in RADIUS Auth/Accounting Requests. WLC supports different format of “Called-Station-ID”. However, for ULI functionality to work, “Called-Station-ID” AVP should be received in AP Identifier format. If Called-Station-ID is received in AP Identifier format then it is sent to GGSN in ULI IE of CPC request.

The “User Location Information” IE is encoded in Cell Global Identifier (CGI) format to indicate WLAN AP location information where the UE is currently located.

The “Geographic Location Type” field is used to convey what type of location information is present in the “Geographic Location” field. To indicate Cell Global Identity format, it should be set to 0.

The “Geographic Location” field is used to convey the actual geographic information as indicated in the “Geographic Location Type” field. The MCC MNC octets should be set to PLMN ID of the PLMN where D-eWAG is located. The LAC and CI octets should be set to Called-Station-ID AP-Identifier LAC and CI components.

After the UE moves to a different access point, WLC sends a RADIUS Accounting Interim with the new Access Point location in “Called-Station-ID” AVP. D-eWAG checks the older ULI and if it is different, it will send UPC Request with ULI with the new Access Point location.

UE MAC to IMEI Mapping Support

The UE MAC to IMEI Mapping Support feature allows user identity information to be provided to the GGSN. This support can be enabled/disabled from the CLI. When enabled, the UE MAC received in “Calling-Station-Id” RADIUS attribute is mapped to IMEIsV and sent in GTP CPC message to the GGSN.

Data-Plane

Uplink Data Path

The uplink data packet from UE is sent by WLC to D-eWAG over the mapped VLAN for that UE session. D-eWAG identifies the session for the received data-packet based on the source IP address. After the session is identified, the data packet is placed over the GTP-U tunnel created with GGSN for this session. This ensures that the packet reaches the appropriate APN network.

Downlink Data Path

D-eWAG uniquely identifies a session based on the GTP-U tunnel from GGSN and extracts the IP packet from GTP-U tunnel. This IP packet contains the destination IP address set to the UE's IP address allocated during DHCP signaling (and actually allocated by GGSN). D-eWAG sends the IP packet downstream to the WLC over the correct VLAN. D-eWAG always uses the same VLAN over which the DHCP packets are received for this UE session in this case. The WLC also takes care of delivering the IP packet to the UE over WLAN.

Overlapping IP Address Support

IMPORTANT:

In this release, Overlapping IP Address support is not fully qualified and is not supported, it is available only for lab testing purposes.

If the IP address allocated by GGSN during the PDP Context Creation is expected to be unique for each UE session (across the different APN/PLMN), then Overlapping IP Address support is not required. In that case, identification of the session for the data-traffic at D-eWAG can be based only on the Source IP address.

To support Overlapping IP addresses, identification of data-traffic is done based on the {VLAN-ID, Source-IP-Address} pair, which ensures that the overlapping IP addresses can exist across operators/APN.

Following table shows the overlapping IP address support in various possible deployment models of D-eWAG:


Table 1. Overlapping IP Address Support
Model Overlapping IP Support Notes

One SSID mapped to one VLAN mapped to one APN.

Yes, the VLAN has to be always different for different APN.

a. Each SSID should always be mapped to unique VLAN in this case even if it is served using multiple WLCs.

b. Different VLAN used for all UE sessions connecting through different SSIDs and uplink packets can be identified uniquely with {vlan+src.ip} pair at D-eWAG.

One SSID mapped to one VLAN mapped to multiple APN.

No

a. Each SSID should always be mapped to a unique VLAN in this case, even if it is served using multiple WLCs.

b. Same VLAN used for all UE sessions and uplink packets cannot be identified uniquely with {vlan+src.ip} pair at D-eWAG.

One SSID mapped to multiple VLAN mapped to one APN.

Yes, the VLAN has to be always different for different APN.

a. WLCs can be different with different VLAN for same SSID.

b. WLC can be configured with AP-Group to use different VLAN.

c. Set of VLANs serving one APN are different from set of VLANs serving another APN. This way overlapping-ip can be supported.



Local Traffic Breakout

The Local Traffic Breakout feature enables the D-eWAG to forward data that does not require 3G access directly to the Internet. With Local Traffic Breakout support the traffic carried by UE will fall into one of the following categories:

  • WLAN Direct IP Access: Carries part of the traffic that will go directly over the Internet. The Gn’ interface is bypassed.
  • WLAN 3GPP IP Access: Carries the 3G traffic that will go in the GTPU tunnel towards the MPC (GGSN).

D-eWAG acts as the AAA proxy as well as DHCP server to the UE attaching to the WLAN network. While acting as DHCP server, D-eWAG creates the PDP context with the GGSN to obtain the IP address to be allocated to the UE through DHCP-Response in the access-side. After the session is created, data is allowed to go through the MPC or directly over the Internet.


Figure 2. D-eWAG with Local Traffic Breakout Deployment

IMPORTANT:

For Local Traffic Breakout support, D-eWAG requires Dynamic NAT functionality for which the ECS and NAT in-line service licenses are required.

APN Selection

A single APN is used for both 3G access and direct IP access. If Local Traffic Breakout is enabled, WLAN subscribers can simultaneously access 3G services and direct IP services.

IP Address Allocation

A WLAN subscriber is always associated with a single IP address, there is no distinction between the Wi-Fi IP address and PDP IP address.

Note that NAT is applied to direct IP traffic, the subscriber’s IP address is NATd and sent to the Internet. In the downlink direction, the destination IP address is changed from the NATd IP address to the subscriber’s IP address and then forwarded to the subscriber.

Controlling Local Traffic Breakout

D-eWAG enables Local Traffic Breakout (direct IP access) based on the availability of Firewall-and-NAT policy for the subscriber. If NAT is enabled for the subscriber then Local Traffic Breakout is enabled.

NAT In-line Service Support

NAT in-line service is required for Local Traffic Breakout support. Local Traffic Breakout is applied to subscriber traffic based on the L3/L4 characteristics—source IP address, source port number, destination IP address, destination port number, and the protocol. One-to-one NAT is applied only for direct IP data while the rest of the 3G data is bypassed by NAT. This can be configured with the help of target-based NAT support. If NAT is enabled, all subscriber IP is NATd. Private IP check of subscriber IP is bypassed.

If NAT is not enabled then all the user data goes to the GGSN.

IMPORTANT:

For D-eWAG, irrespective of the NAT pool type, NAT IP address is allocated only on demand—after the data requiring NAT comes in.

Enabling Firewall-and-NAT Policy

The Firewall-and-NAT policy can be enabled for a subscriber in one of the following ways:

  • Subscriber Template
  • RADIUS AVP
  • ECS Rulebase

The Firewall-and-NAT policy can either be specified in the ECS rulebase, which can in turn be specified in the Subscriber Template, or the policy can be specified directly in the Subscriber Template.

Subscriber configuration has higher priority compared to the ECS rulebase configuration. Therefore, if Firewall-and-NAT policies are configured both in the Subscriber Template and in the ECS rulebase, the policy specified in the Subscriber Template is applied for the subscriber.

Target-based NAT Configuration

A NAT Realm (NAT IP Pool from where the NAT IP can be assigned to a subscriber) can be selected based on the L3/L4 characteristics of the flows / connections coming from the subscriber.

This association is done with the help of Access rules configurations in the rulebase. The administrator can configure the realm names along with the Access rules in the Firewall-and-NAT policy. The matching criteria for these rules in the rulebase can be based on the L3/L4 parameter. This allows the realms to be selected based on L3/L4 parameters of the flow (target-based NAT). When packets matching a given ruledef r1 are received, NAT is done using the NAT IP address allocated to the subscriber from the realm configured for the ruledef r1. In this way, the NAT realm/NAT IP address to be used for subscriber flows is decided during rule match.

If no NAT realm name is found in the ruledef matching the packet, or if it is specified to bypass NAT, NAT will not be applied on the subscriber flow. The traffic is routed within the private network.

Thus for NAT to be applied, a realm name must be configured in the matching ruledef. If NAT has to be bypassed, then a NAT realm must not be configured in the ruledef.

Data Path Flow

In the uplink direction, irrespective of the data received at D-eWAG, D-eWAG will apply the ACS ruledef specified. For 3G data, as per the ruledef configuration NAT will be bypassed. For direct IP data, NAT is applied to the destination address. After the ACS is processed NAT status will decide whether the data should directly go over the Internet or in the GTPU tunnel towards the GGSN.

In the downlink direction, MPC data received at the SGTP interface in GTPU tunnel goes directly towards the UE. While the data from direct IP connection received at D-eWAG is NATd and sent to the UE.

IMPORTANT:

Note that NAT is applied only for the direct IP data based on the access rules defined.

Data Path Changes

When using WLAN direct IP access, a WLAN UE has to use its local IP address. As the WLAN local IP address and the GGSN assigned IP address are same, NAT support is required for direct IP access. All the traffic between WLAN UE and direct IP connection is NATd.

Uplink Data Path

All 3G service data is NAT bypassed while other direct IP data is NATd. After ECS and NAT processing is done, if flagged, the data is sent directly over the Internet. Else, the data is sent to the GGSN over the GTPU tunnel.


Figure 3. Uplink Data Path

Downlink Data Path

Data from 3G services is received in GTPU tunnel while the NATd data from Internet is received directly. In the downlink data path, after ECS processing is done the data is sent to the UE.


Figure 4. Downlink Data Path

Recovery Support

The NAT framework takes care of recovering the NAT status and NAT flow. For the Local Traffic Breakout counters, new micro checkpoint is added, which is sent as part of clp stats for D-eWAG callline.

Accounting Support

Direct IP data is accounted separately.

The following RADIUS AVPs support direct IP counts:

  • SN-LBO-Acct-IN-Pkts: Indicates number of packets sent by UE directly to the Internet
  • SN-LBO-Acct-Out-Pkts: Indicates number of packets received by UE directly from the Internet.
  • SN-LBO-Acct-IN-Octets: Indicates number of octets sent by UE directly to the Internet.
  • SN-LBO-Acct-Out-Octets: Indicates number of octets received by UE directly from the Internet.

Note that whereas direct IP data is accounted separately, there is only a cumulative Total Uplink and Total Downlink data count available for the UE. It is not possible to identify 3G data sent for the subscriber from accounting messages or CDR.

Differentiated Services Code Point Marking

Differentiated Services Code Point (DSCP) levels can be assigned to specific traffic patterns in order to ensure that data packets are delivered according to the precedence with which they are tagged. The DiffServ markings are applied to the IP header of every subscriber data packet transmitted in the downlink and/or uplink direction based on negotiated QoS at GGSN and local configuration in the IPSG service.

DSCP values must be configured for different QCI values. The following table presents the traffic class to QCI mapping (based on 3GPP spec 23.203).


Table 2. Traffic Class to QCI Mapping
GPRS QoS Class Identifier Value UMTS QoS Parameters
Traffic Class THP Signalling Indication Source Statistics Descriptor

1

Conversational

N/A

N/A

speech

2

Conversational

N/A

N/A

unknown

3

Streaming

N/A

N/A

speech

4

Streaming

N/A

N/A

unknown

5

Interactive

1

Yes

N/A

6

Interactive

1

No

N/A

7

Interactive

2

No

N/A

8

Interactive

3

No

N/A

9

Background

N/A

N/A

N/A



For the downlink path, DSCP markings can be configured to control the DSCP markings for downlink packets. IP header of the packet is updated with value in the TOS field.

For uplink traffic—traffic from D-eWAG to GGSN through GTP tunnel—DSCP markings can be configured. In this case, only outer IP header is used for routing the packet over Gn' interface. Hence, TOS field of only outer IP header is changed, that is subscriber packet is not marked with DSCP value at D-eWAG.

DSCP marking can be configured with a “pass through option”, which when configured uses the marking received on ingress to mark packets on egress.

IMPORTANT:

Note that Traffic Policing/Shaping is not supported in this release.

Bulk Statistics Support

The system's support for bulk statistics allows operators to choose to view not only statistics that are of importance to them, but also to configure the format in which it is presented. This simplifies the post-processing of statistical data since it can be formatted to be parsed by external, back-end processors.

When used in conjunction with the Web Element Manager, the data can be parsed, archived, and graphed.

The system can be configured to collect bulk statistics (performance data) and send them to a collection server (called a receiver). Bulk statistics are statistics that are collected in a group. The individual statistics are grouped by schema.

For the list of supported schema and information on how to configure them, refer to the DHCP-based Enhanced Wireless Access Gateway Configuration chapter.

The system supports the configuration of up to four sets (primary/secondary) of receivers. Each set can be configured with to collect specific sets of statistics from the various schema. Statistics can be pulled manually from the system or sent at configured intervals. The bulk statistics are stored on the receiver(s) in files.

The format of the bulk statistic data files can are configurable, operators can specify the format of the file name, file headers, and/or footers to include information such as the date, system host name, system uptime, the IP address of the system generating the statistics (available for only for headers and footers), and/or the time that the file was generated.

When the Web Element Manager is used as the receiver, it is capable of further processing the statistics data through XML parsing, archiving, and graphing.

The Bulk Statistics Server component of the Web Element Manager parses collected statistics and stores the information in the PostgreSQL database. If XML file generation and transfer is required, this element generates the XML output and can send it to a Northbound NMS or an alternate bulk statistics server for further processing.

Additionally, if archiving of the collected statistics is desired, the Bulk Statistics server writes the files to an alternative directory on the server. A specific directory can be configured by the administrative subscriber or the default directory can be used. Regardless, the directory can be on a local file system or on an NFS-mounted file system on the Web Element Manager server.

IMPORTANT:

For more information on bulk statistics configuration, refer to the Configuring and Maintaining Bulk Statistics chapter in the System Administration Guide.

Threshold Crossing Alerts Support

Thresholding on the system is used to monitor the system for conditions that could potentially cause errors or outage. Typically, these conditions are temporary (i.e. high CPU utilization, or packet collisions on a network) and are quickly resolved. However, continuous or large numbers of these error conditions within a specific time interval may be indicative of larger, more severe issues. The purpose of thresholding is to help identify potentially severe conditions so that immediate action can be taken to minimize and/or avoid system downtime.

The ASR5000 chassis supports several threshold values of which the following are applicable to D-eWAG:

  • Call setup:
    • Number of calls setup
  • Subscriber number:
    • Total number
  • Licensed session utilization
  • Port utilization:
    • High activity
    • Transmit utilization
    • Receive utilization
  • PAC/PSC CPU resource availability:
    • Percent utilization
    • Available memory
    • Load
    • Memory usage
    • Session throughput
  • SPC/SMC CPU resource availability:
    • Memory usage
    • Percent utilization
  • Packet processing:
    • Number of packets filtered/dropped
    • Number of packets forwarded to CPU

Note that the other thresholds are platform specific and so are applicable to D-eWAG as well.

The following thresholding models are supported by the system:

  • Alert: A value is monitored and an alert condition occurs when the value reaches or exceeds the configured high threshold within the specified polling interval. The alert is generated then generated and/or sent at the end of the polling interval.
  • Alarm: Both high and low threshold are defined for a value. An alarm condition occurs when the value reaches or exceeds the configured high threshold within the specified polling interval. The alert is generated then generated and/or sent at the end of the polling interval.
Thresholding reports conditions using one of the following mechanisms:
  • SNMP traps: SNMP traps have been created that indicate the condition (high threshold crossing and/or clear) of each of the monitored values.
    Generation of specific traps can be enabled or disabled on the chassis. Ensuring that only important faults get displayed. SNMP traps are supported in both Alert and Alarm modes.
  • Logs: The system provides a facility called threshold for which active and event logs can be generated. As with other system facilities, logs are generated. Log messages pertaining to the condition of a monitored value are generated with a severity level of WARNING.
    Logs are supported in both the Alert and the Alarm models.
  • Alarm System: High threshold alarms generated within the specified polling interval are considered “outstanding” until a the condition no longer exists or a condition clear alarm is generated. “Outstanding” alarms are reported to the system's alarm subsystem and are viewable through the Alarm Management menu in the Web Element Manager.
    The Alarm System is used only in conjunction with the Alarm model.

IMPORTANT:

For more information on thresholds, refer to the Thresholding Configuration Guide.

Congestion Control Support

IMPORTANT:

In this release, Congestion Control support is not qualified and is not supported, it is available only for lab testing purposes.

The Congestion Control feature enables to specify how the system reacts in a heavy load condition. Congestion control operation is based on configuring congestion condition thresholds and service congestion policies.

IMPORTANT:

Overload Disconnect is not supported.

Congestion Control monitors the system for conditions that could potentially degrade performance when the system is under heavy load. Typically, these conditions are temporary (for example, high CPU or memory utilization) and are quickly resolved. However, continuous or large numbers of these conditions within a specific time interval may have an impact on the system’s ability to service subscriber sessions. Congestion control helps identify such conditions and invokes policies for addressing the situation.

Congestion control operation is based on configuring the following:

  • Congestion Condition Thresholds: Thresholds dictate the conditions for which congestion control is enabled and establishes limits for defining the state of the system (congested or clear). These thresholds function in a way similar to operation thresholds that are configured for the system as described in the Thresholding Configuration Guide. The primary difference is that when congestion thresholds are reached, a service congestion policy and an SNMP trap are generated.
    A threshold tolerance dictates the percentage under the configured threshold that must be reached in order for the condition to be cleared. An SNMP trap is then triggered.
    • Port Utilization Thresholds: Congestion thresholds for utilization of all ports in the system.
      • Port-specific Thresholds: Congestion thresholds for individual ports.
    • Service Congestion Policies: Congestion policies are configurable for each service. These policies dictate how services respond when the system detects that a congestion condition threshold has been crossed.
    • License Utilization: Congestion thresholds for license utilization on the system.
    • Maximum Sessions-per-Service Utilization: Congestion thresholds for maximum number of sessions allowed per service.

IMPORTANT:

For more information on the Congestion Control feature, refer to the Congestion Control chapter in the System Administration Guide.

Redundancy Support

IMPORTANT:

In this release, D-eWAG supports basic Session Recovery, ICSR is not supported.

IMPORTANT:

In this release Line Card Switchover is not supported.

Session Recovery feature provides a mechanism to recover failed Session Manager (SessMgr) task(s) without any call loss. Recovery framework is same as used by other products. A minimum of four PSCs (three active and one standby) is required in an ASR5000 chassis to support the Session Recovery feature. This is because the DEMUX Manager and VPN Manager tasks run on a PSC where no SessMgr runs when session recovery is enabled and one PSC is used as standby PSC. The other two PSCs run SessMgr and AAAMgr tasks.

Session Recovery is a licensed feature and can be controlled from the CLI, that is enabled/disabled Session Recovery across the whole chassis. When the CLI is used to configure the Session Recovery feature, Session Controller updates each SessMgr task.

In the case of D-eWAG, the IPSG Manager, SGTPC Manager, and VPN Manager run on one PSC. SessMgr runs on one separate PSC. AAAMgr runs on one separate PSC and on one standby PSC. Therefore, a minimum of four PSCs (three active and one standby) are required.

For D-eWAG Session Recovery support, apart from common access-side attributes (common between D-eWAG and R-eWAG sessions), attributes specific to D-eWAG session such as Default-GW-IP address, UE-MAC, and so on are supported. D-eWAG GTP context information is recovered similar to R-eWAG as Gn' interface is used by both.

Charging

User traffic towards mobile packet core is accounted by GGSN in collaboration with existing 3G Charging Gateway Function. D-eWAG supports the following accounting for the user-traffic:

  • RADIUS accounting
  • GTPP accounting (CDR)

Offline Charging

In Offline Charging, charging information is collected concurrently with resource usage. The charging information is then passed through a chain of logical charging functions, and the CDR files are generated by the network, which are then transferred to the network operator's Billing Domain.

The CTF (an integrated component in each charging relevant NE) generates charging events and forwards them to the CDF. The CDF, in turn generate S-CDRs, which are then transferred to the CGF. Finally, the CGF create S-CDR files and forwards them to the Billing Domain. The CTF and CDF are integrated in the D-eWAG. However, the CGF may exist as a physically separate entity or integrated to the D-eWAG. If the CGF is external to the D-eWAG, then the CDF forwards the CDRs to the CGF across the Gz/Wz interface (using GTPP protocol).

In the ASR5000 chassis, D-eWAG is integrated with the CTF and CDF functions and it generates S-CDR based on the triggered events and sends the same to the CGF over the Gz/Wz interface. Note that S-CDR is used by SGSN, and the same format is used for D-eWAG.

The D-eWAG Offline charging involves the following functionalities for WLAN 3GPP IP Access:

  • Charging Trigger Function
  • Charging Data Function
  • Gz/Wz Reference Point

Triggers for Charging Information Addition and CDR Closure

D-eWAG uses the Charging Characteristics to determine whether to activate or deactivate CDR generation. The Charging Characteristics are also used to set the coherent chargeable event conditions (for example, time/volume limits that trigger CDR generation or information addition). Multiple Charging Characteristics “profiles” may be configured in the D-eWAG to allow different sets of trigger values.

Triggers for S-CDR Closure

The following events trigger closure and sending of a partial S-CDR:

  • Time Trigger (every x seconds configured using “interval x”)
  • Volume Trigger (every x octets configured using “volume x” (up/down/total))
  • On reaching maximum number of container limit
  • Command gtpp interim now

An S-CDR is closed as the final record of a session for the following events:

  • UE-initiated call termination
  • Admin release at D-eWAG via clear sub all
  • GGSN-initiated call termination
  • Abnormal releases due to multiple software failures.
  • UE-initiated DHCP release
  • AAA-initiated call disconnect
  • WLC-initiated call termination

Triggers for S-CDR Charging Information Addition

The “List of Traffic Volumes” attribute of the S-CDR consists of a set of containers, which are added when specific trigger conditions are met, and identify the volume count per PDP context, separated for uplink and downlink traffic, on encountering that trigger condition.

Billing Record Transfer

The S-CDR generated can either be stored on Hard Disk (GSS) or can be transferred to the CGF. Local storage is also available. Gz/Wz is the offline charging interface (CDR-based) between the GSN and the CGF. The D-eWAG supports both GSS and GTPP-based record transfer.

Lawful Intercept Support

The Lawful Intercept (LI) functionality provides network operators the ability to intercept control and data messages of suspicious subscribers. The ASR5000 chassis provides a proprietary interface to third-party Mediation Function (MF) or Delivery Function (DF), and supports LI for D-eWAG.

For more information on LI support, contact your accounts representative.

D-eWAG + R-eWAG Combo Deployment

IMPORTANT:

In this release, the D-eWAG + R-eWAG combo deployment option is not qualified and is not supported, it is available only for lab testing purposes.

The D-eWAG and R-eWAG services can be deployed on the same chassis. This is possible because R-eWAG operates based on APN profile and D-eWAG operates based on subscriber-template. This clearly separates the user profile selection process for these services without affecting each others configurations.

The only known restriction is that both these services cannot be configured in the same context. Also, note that the context-replacement issue at GGSN due to same IMSI+NSAPI will not be the issue in R-eWAG + D-eWAG combo setup as the UE can attach to only one WLAN at a time. Thus, it cannot connect through both R-eWAG and D-eWAG at the same time.

IMPORTANT:

In this release, NAT policy must not be configured for D-eWAG. In D-eWAG + R-eWAG combo deployments NAT is required for R-eWAG, it must be ensured that NAT policy is not configured for D-eWAG ECS session.

How it Works

The following illustration shows network setup for the D-eWAG-based solution for MPC access.


Figure 5. D-eWAG Network Setup

This section presents call procedure flows for the following scenarios:

Session Setup

This section presents the call flow for session setup scenario.


Figure 6. D-eWAG Session Setup Call Flow

Figure 7. D-eWAG Session Setup Call Flow... continued

Figure 8. D-eWAG Session Setup Call Flow... continued

Session Teardown

This section presents call flows for session teardown scenarios.

Session Teardown - AAA Initiated

This section presents the call flow for AAA-initiated Session Teardown scenario.


Figure 9. Session Teardown - AAA Initiated Call Flow

Session Teardown - GGSN Initiated

This section presents the call flow for GGSN-initiated Session Teardown scenario.


Figure 10. Session Teardown - GGSN Initiated Call Flow

Session Teardown - UE Initiated

This section presents the call flow for UE-initiated Session Teardown scenario.


Figure 11. Session Teardown - UE Initiated Call Flow

Session Teardown - WLC Initiated

This section presents the call flow for WLC-initiated Session Teardown scenario.


Figure 12. Session Teardown - WLC Initiated Call Flow

Session Update

This section presents call flows for session update scenarios.

Session Update - AAA Initiated

This section presents call flow for the AAA-initiated session update scenario.


Figure 13. Session Update - AAA Initiated Call Flow

Session Update - GGSN Initiated

This section presents call flow for the GGSN-initiated session update scenario.

GGSN-initiated UPC Request for QoS update would be processed at D-eWAG and the QoS associated with the session would be updated. UPC request for update of any other parameter would be rejected by D-eWAG. The GGSN may initiate a DPC because of this.

IMPORTANT:

Note that D-eWAG internally uses R7-QoS regardless of which QoS is requested and negotiated. When D-eWAG receives UPC from GGSN, it compares it with QoS requested by AAA, and QoS with smaller version is selected for UPC response. In case of same version, QoS with small Maximum Bit Rate (MBR) is selected.

IMPORTANT:

In this release, D-eWAG does not generate CoA RADIUS Request to WLC.


Figure 14. Session Update - GGSN Initiated Call Flow

Session Update - WLC Initiated

This section presents call flow for the WLC-initiated session update scenario.

WLC cannot send 3gpp-qos. Thus the UPC from D-eWAG to GGSN for QoS change from WLC will not happen. UPC will only be sent for AP information change.


Figure 15. Session Update - WLC Initiated Call Flow

Dependencies and Limitations

This section lists limitations to the D-eWAG solution in this release.

  • IPSG-Service Configuration Restriction: Only one IPSG service must be configured per context. Multiple IPSG services must not be configured in the same context as the IPSG will not be able to differentiate between uplink and downlink packets.

Deployment Models

  • General assumptions:
    • D-eWAG acts as first-hop L3 router for WLC.
    • WLC and D-eWAG nodes are in a centralized location and the connectivity between WLC and D-eWAG is based on VLANs.
    • D-eWAG acts as authentication-proxy and dhcp-server for all the 3G-SSID being served by WLC.
    • Data path follows flexconnect (or H-REAP) model in Wi-Fi access with the data forwarded to the controller in the centralized location from all the APs.
    • The RADIUS control path and {dhcp+datapath} can be in the same or different VLAN between WLC and D-eWAG depending on whether the AP-Group or AAA-VLAN-Override feature is enabled or disabled.
      Note that enabling the AP-Group at WLC or AAA-VLAN-Override feature at 3GPP-AAA server will not affect D-eWAG functionality as Overlapping IP address is not supported in this release.
    • This D-eWAG solution is not tested against non-Cisco WLC nodes. However, it should work with WLCs from any vendors as long as it satisfies the requirements mentioned in the Requirements in WLC section, though it is advised to use Cisco WLC nodes for better interoperability.
  • Assumptions in uplink:
    • UE default-router (gateway) configuration is provided by ASR5000 chassis in DHCP Response with IP address in the same subnet as the UE IP address.
    • WLC forwards the ARP requests for default-gateway from UE to the appropriate ASR5000 chassis VLAN interface.
    • ASR5000 chassis responds to default-gateway ARP requests with MAC address of the VLAN interface on which it is received.
    • UE uses the ASR5000 chassis returned MAC addresses as destination MAC for uplink data packets.
    • ASR5000 chassis identifies the session for the data packet based on Source-IP flow.
    • Overlapping IP address is not supported in this release.
  • Assumptions for downlink:
    • D-eWAG is aware of the VLAN mapped to data path of the UE-Session. This is the same VLAN used for communication of DHCP-signaling between WLC and D-eWAG.
    • D-eWAG sends the data packet with Dest-IP as UE-IP and Dest-MAC as UE-MAC to WLC on the data VLAN.

Requirements in WLC

WLC node capabilities for D-eWAG service:

  • Each 3G-SSID served by WLC should be mapped to a VLAN. The default-gateway for this VLAN should be configured with the IP address of the corresponding VLAN interface at D-eWAG so that any signaling packet generated for that SSID (like RADIUS Access-Request, DHCP, etc) will reach D-eWAG.
  • RADIUS server IP address for the 3G-SSID at WLC should be configured with the D-eWAG service IP address. This is required since D-eWAG acts as RADIUS Proxy and inspects the authentication exchanges between UE and 3GPP-AAA for obtaining the 3G attributes required to create PDP-context with the GGSN.
  • DHCP server IP address for the 3G-SSID at WLC should be configured with the D-eWAG service IP address. This is required since D-eWAG acts as DHCP-server and notifies the IP address allocated by GGSN using DHCP signaling to the UE.
  • WLC should be configured to use its VLAN interface's IP as Source-IP for RADIUS/DHCP Relay signaling packets and not the management interface IP. This is required since D-eWAG verifies the shared-secret for RADIUS communication based on this IP address.

Requirements at GGSN

The IP-Pool subnet range configured at GGSN for APN network access should have one IP-address dedicated as default-gateway address for that subnet and not allocated to any UE. This IP address should be configured at D-eWAG node as default-gateway IP address for that APN.

For example, if the IP-pool subnet range for an APN is 12.0.0.1 to 12.0.0.100, then one IP-address from this range, say 12.0.0.1 is dedicated as default-gateway address for this subnet range. Thus, the GGSN IP-pool configuration in the D-eWAG-based solution should be changed to { 12.0.0.2 to 12.0.0.100 } range and the IP-address 12.0.0.1 is configured at D-eWAG node. Also, 12.0.0.1 is conveyed to UE as default-gateway during DHCP-Offer message by D-eWAG so that it acts as default-gateway for all the uplink data-packets from the UE.