Mobile MPLS and Its Key Technologies

Mobile data is one of the most promising Telecom businesses. It can meet the needs of various mobile terminal users to access the Internet anytime, anywhere, it can also meet the needs of data and multimedia communication between them. Mobile IP is one of the key technologies to implement mobile data services. It supports mobile terminals with fixed IP addresses to move freely on the Internet and keep communication uninterrupted during the mobile process. Multi-Protocol Label Switching (MPLS) is a Fast Packet Switching technology that combines layer-3 IP routing technology and layer-2 switching technology. It classifies data at the edge of the network, tags are added for each type of data, and only tags are exchanged in the network, which enables Fast Packet Exchange. MPLS overcomes the need for traditional IP networks to provide best effort? Service disadvantages, can ensure the user's QoS requirements, and improve the utilization of network resources through application traffic engineering. To provide high-quality mobile data services with QoS Assurance, mobile MPLS technology emerged. It combines Mobile IP addresses and MPLS networks to support IP Mobility in MPLS networks, which not only meets users' mobility requirements, in addition, it can provide good guarantees in terms of user service quality and full use of network resources, and has great market potential and business prospects.

I. Mobile MPLS Principle

Basic MPLS does not support mobility management. The mobile MPLS network formed by the combination of Mobile IP and MPLS enables the MPLS network to support IP mobility. There are currently two kinds of mobile MPLS technology: basic mobile MPLS and Hierarchical Mobile MPLS (H-MPLS ). Basic mobile MPLS only takes into account the relatively slow movement of IP users in a large range. When the use of micro-cells in the wireless access network causes frequent IP users to switch between subnets, the switching latency is large and may cause communication interruption. Hierarchical Mobile MPLS overcomes the disadvantages of basic mobile MPLS and has the ability to manage micro-cellular mobility.

1. Basic mobile MPLS

In the basic mobile MPLS mechanism, when a mobile host MH enters a non-home subnet, a registration message is sent to the foreign proxy FA of the subnet ), FA forwards this registration message to the hometown proxy HA of MH through a normal IP address route), and HA receives the registration message and obtains the MH transfer address COA ). COA can be the IP address of FA or a temporary IP address that FA dynamically assigns to MH. Then HA sends the tag request message of the Label Distribution Protocol LDP to FA, and FA sends the tag ing message of LDP back to HA. When the tag ing message arrives at HA, the label exchange path LSP between HA and FA is established. Then, HA finds the row that uses the hometown address of MH as the Forwarding Equivalence Class FEC by searching the tag table, change the outgoing port and outgoing label of the row to the port and label value used by the LSP between HA and FA. At last, HA sends the registration Response Message to FA through the established LSP between HA and FA. After receiving the registration response message, FA adds a row to its tag table and fills in the received tag value and port value in the incoming tag domain and inbound port domain of the row.

If a communication node CN sends data packets to MH, the data packets are routed to the home network of MH and intercepted by HA, as described in Mobile IP protocol. As the Label Edge Router LER of the LSP entry), HA will search for the packet's outgoing label and outgoing port in the label table based on the IP address contained in the received packet. Based on the found outgoing label and outgoing port, the packet is tagged and transmitted along the LSP between HA and FA in tag exchange mode. After receiving the data packet, the FA will find its label table. Because FA is the exit of the LSP, the corresponding outgoing label and outbound port in the label table are empty ). As the LSP egress LER, the FA removes the packet label and sends the packet to MH through the IP layer. At this point, MH receives the packet sent from the communication node. In basic mobile MPLS, when MH moves to a new MPLS subnet, it sends a registration message to its HA through the FA of The New subnet, create a new LSP between the new FA and HA. The time required to create a new LSP and transfer data to the new LSP is called the switching latency. If the LSP switching latency is large, data communication between the communication node and MH will be interrupted, resulting in data loss. In order to reduce the switching latency and ensure communication during the switching process, Hierarchical Mobile MPLS is introduced.

2. Hierarchical Mobile MPLS

Hierarchical Mobile MPLS technology assumes that an MPLS network is divided into multiple MPLS domains, and each MPLS domain has multiple access subnets. In addition, Hierarchical Mobile MPLS also introduces a new component, that is, the external region proxy FDA), each MPLS domain has an FDA. In the Hierarchical Mobile MPLS mechanism, MH can determine whether it is in the home network or a foreign network based on the proxy broadcast message received. If MH determines that it is in a foreign network, it will request a COA from FA and send the registration request message to FA. FA forwards the registration request message to the FDA in this MPLS domain, instead of the home site proxy of MH. If the first time the MH was moved to this MPLS domain, the FDA will forward the registration message to the hometown proxy of MH. After obtaining the registration message and obtaining the fda ip address, HA sends the tag request message to the FDA using LDP. After receiving the tag request message, the FDA sends the tag ing message to the HA and sends the tag request message to the FA of the subnet in the current MPLS domain of MH. After the tag ing message arrives at HA, the LSP between FDA and HA is established. Similarly, FA sends a tag ing message to the FDA. When the tag ing message arrives at the FDA, the LSP between the FDA and FA is also established. Then, HA finds the row in the label table with the MH hometown address as FEC, and changes the outbound port and outbound label to the corresponding LSP value between the FDA and HA. Finally, HA sends a registration Response Message to the FDA along its LSP to the FDA, and the FDA also forwards the registration response message along the FDA to the fa lsp. After receiving the registration response message, FA adds a new row to its tag table and enters the tag value and port number of the received registration response message in the incoming tag domain and inbound port domain. Like basic mobile MPLS, HA intercepts data packets when a communication node sends data packets to MH located in a foreign network in the Hierarchical Mobile MPLS mechanism. In a Hierarchical Mobile MPLS network, HA looks for its label table and finds the outgoing label and port corresponding to the packet. Based on the found results, the data packet is sent to the FDA by the HA with labels along the LSP between the HA and the FDA. After receiving this packet, the FDA will continue to find the corresponding outgoing tag value based on the incoming tag value, that is, forward the packet to FA along the LSP between the FDA and FA. After receiving the packet, FA queries its label table. Because FA is the exit of the LSP, both the outbound port and the outbound tag value in the label table are empty ). Finally, FA removes the label on the data packet and forwards the data packet to MH through the IP layer. Therefore, MH receives the data sent from the communication node.

If MH switches from one subnet to another in the same MPLS domain, it will request a new COA for the FA in The New subnet and send a registration request message to the FA in The New subnet. The FA in The New subnet will forward the registration request message to the FDA. After receiving the registration request message, the FDA sends a tag request message to the FA in The New subnet, and then the FA in The New subnet sends the tag ing message to the FDA, in this way, a new LSP is established between the FA and FDA of The New subnet, but the LSP between HA and FDA has not changed. Obviously, compared with basic mobile MPLS, Hierarchical Mobile MPLS does not need to establish a brand new LSP between HA and the FA of The New subnet, but only creates a new LSP between the FDA and the FA of The New subnet, this greatly reduces the switching latency. During the switchover, MH can also send a binding update message to notify the original FA of its new COA. The old subnet FA can cache the new binding relationship of MH. If the FDA uses an outdated label table to send data packets to MH, the old subnet FA will establish an LSP with the new FA and send data packets to the new FA through the LSP, this avoids packet loss during the switching process.

In the Hierarchical Mobile MPLS mechanism, the movement of MH in the same MPLS domain is transparent to its HA. In this way, the switching latency is reduced, and the message interaction required for switching across the network is reduced, thus saving network resources.

II. Key Technologies of mobile MPLS

According to the principle of mobile MPLS, mobile MPLS should solve three problems: Switching delay and communication interruption, triangle routing, and QoS. 1. The goal of the quick switch technology is to minimize the switching latency and ensure that the communication is not interrupted during the switching process, that is, no packet loss occurs during the switching process ). Because re-registration is required between MH, new FA, New FDA, and HA during the switching process, the simpler the registration mechanism, the lower the switching latency. At the same time, the shorter the time to create a new LSP between MH and HA, the less likely the data packet loss will be during the switching process.

1) the basic idea of establishing the LSP mechanism before the switchover is to create a new LSP before the MH moves to a new foreign Subnet or a foreign region. In Hierarchical Mobile MPLS, if MH moves in the adjacent subnet, LSP is established in the adjacent subnet before MH moves to the adjacent subnet. If MH moves between the adjacent subnets, then, the LSP is established in the phase neighborhood before the MH is moved to the phase neighborhood. The active LSP and passive LSP are introduced in this mechanism. The active LSP refers to the LSP that is being used to transmit data from the FDA to the FA in the field where the MH is currently located, passive LSP refers to the unused LSP of FA from the FDA to all adjacent subnets in the foreign subnet where MH is currently located. If MH knows the subnet to be accessed, it can use the FA Discovery Protocol to obtain the IP address of the FA in the adjacent subnet, and then send the access request message with the access time to the FA, after receiving the access request message, these FA can establish a passive LSP with the FDA. The access time is the time that MH estimates to move from the current subnet to a subnet. If MH does not send any access request to refresh the message before the access time-out and does not move it to this subnet, the corresponding passive LSP entries in the label table will be deleted. After MH is moved to a new subnet, it will send an update message to the FA in the original subnet, and then the FA will send it to the FDA, in this case, the intermediate routers change the LSP state they save from active LSP to passive LSP. At the same time, MH will send an update message to the new FA. After receiving the message, the FA will also send it to the FDA. The intermediate router will change the LSP status they save from passive LSP to active LSP. Before the switchover, the LSP mechanism is established. When the switchover occurs, you only need to use one message to complete the LSP switching. This greatly reduces the latency of MH switching from one subnet to another. This mechanism can be extended to MH switching between different MPLS domains.

2) The multicast-based fast switching mechanism is a fast switching mechanism that establishes LSP between the FDA and the FA in the subnet of the current MH, and between all the FA in the subnet adjacent to the subnet, these lsns form a group. When the FDA receives a request for MH switchover, it broadcasts all data packets from the communication node to the FA of all adjacent subnets through a pre-established LSP. In this way, when MH moves to a new subnet, it can immediately receive data packets from the communication node. After the switchover is complete, the network will also be adjusted based on the information during the switchover process to register the new LSP and delete the unnecessary LSP.

2. triangle routing Solution

In mobile MPLS, data packets sent from communication nodes to MH are first sent to HA and then forwarded by HA along LSP to MH. This is a so-called "triangular routing" problem. Because packet forwarding is not sent to MH through the optimal LSP, but must be forwarded through HA, which wastes network resources and leads to high transmission latency. If the communication node can obtain the COA of the current MH subnet and establish an LSP between the communication node and the FA, data packets do not have to pass through HA, you can directly forward data through the LSP between the communication node and the FA to solve the "triangle routing" problem.

1) COA Cache Mechanism

The COA cache mechanism caches the COA of one or more subnets of MH on the communication nodes. When data on the Communication Node is sent to MH, first find the items in the COA cache table that correspond to MH. If the COA bound to MH is found, the COA is used to establish an LSP between the communication node and the FA in the current subnet of the MH, and send data packets through the LSP between the communication node and the FA, without the need for HA forwarding. The COA cache related to MH does not exist in the communication node at the beginning. The packets sent to MH still need to be forwarded through HA. However, according to the registration process of MH moving between subnets, HA contains the COA information of the subnet where MH is currently located. Therefore, when the data packet arrives at HA, HA will find the COA of the current MH subnet and forward the data through the LSP established between HA and FA. At the same time, HA finds out the COA corresponding to the data sender's address and the data destination address based on the received data packet, and sends the MH and Its COA binding information to the data sender's communication node. After receiving the binding information, the Communication Node caches it and uses the binding information to directly establish LSP with the FA of the current MH subnet. After that, the packets sent from the communication node to MH no longer need HA forwarding, but are directly sent through the LSP between the communication node and FA. The COA information cached by the communication node can be refreshed regularly or by the HA when the MH moves to the communication node to ensure the accuracy of the COA information.

2) feature data profile-based mechanism

The mechanism based on user behavior solves the "triangle routing" problem by predicting the mobile characteristics of MH. Under this mechanism, each MH corresponds to a set of feature data describing user behavior information, including the user's mobility characteristics, travel plans and possible communication requirements. When a communication node needs to send data to MH, it first tries to obtain the mobile feature data of MH and then deduce the COA of the subnet where MH is located, create an LSP between the communication node and the FA through COA, and then send data through LSP. The mechanisms for obtaining MH feature data include distributed and centralized. After the Communication Node obtains the characteristic data of MH, it can infer the COA of the current MH subnet and establish the LSP with FA accordingly. The COA information derived from the MH feature data may not be accurate enough, but it can be improved through the Hierarchical Mobile MPLS mechanism.

3. QoS implementation in Mobile MPLS

MPLS implements QoS through DiffServ and IntServ service models. These two methods are also applicable to mobile MPLS and provide quality assurance for real-time mobile multimedia services.

1) Use DiffServ to provide QoS

In the DiffServ service model, business flows are divided into three types at the edge of the network: speed up forwarding EF), determine forwarding AF) and do what you can. Each type corresponds to a type mark, DSCP ). The core node in the network determines the behavior of each packet (PHB) by checking the DSCP value of the business flow ).

In a mobile MPLS network that uses DiffServ, each MH registers its own business type in its hometown proxy, and the business type information is stored on HA. When MH is moved to the subnet of another domain, its business type information is forwarded to the corresponding domain by HA. Based on the service type of MH, you can establish LSP that meets the QoS requirements of MH service type in the external region. Taking Hierarchical Mobile MPLS as an example, when MH is first moved to an external region, the registration message is sent to HA through the FDA, and HA sends the registration Response Message to the FDA, in this case, HA can send the business type information about MH to the FDA, which stores the information. Obviously, the FDA should store the business type information of all MH roaming in the current domain. When the FDA sends a registration Response Message to FA, it can also send the MH service type information to FA, so that FA can establish a E-LSP with the communication node to realize QoS.

As the entry ler of the E-LSP, FA and communication nodes have MPLS/DiffServ functions. Based on the characteristics of the business flow, they perform timing, shaping, and DSCP value allocation on the data flow, and add tags to the data packets. The intermediate node of the MPLS network, such as the FDA) selects the corresponding PHB Based on the DSCP value of the received data packet to forward the data. In mobile MPLS with DiffServ, the registration information and service type information can be transmitted on the pre-established LSP for transmitting signaling, the business flow can be transmitted between HA and FDA, and between FDA and FA, multiple LSPs that meet the QoS requirements of the business type. Mobile MPLS using DiffServ has two features: first, the service mechanism of nodes inside the network is relatively simple, and internal nodes only perform simple scheduling and forwarding, the storage and monitoring mechanisms of stream status information are only performed on the Boundary Node. Second, the service objects of nodes in the network are stream aggregation rather than single streams. The single stream information is only saved on the network boundary, therefore, it has good scalability and robustness.

2) use IntServ to provide QoS

In the IntServ service model, Resource Reservation Protocol (RSVP) is used to set up a soft state for each stream segment by segment, and accept control is used to determine whether the link or network node has sufficient resources to meet QoS requirements. The two main messages of RSVP are PATH and RESV. The PATH is the message from the sender to the receiver. The RESV message contains the traffic characteristics and classification of each stream. After receiving the PATH Message, the receiver returns the RESV message, which contains QoS requirements. In order to use IntServ to support QoS in Mobile MPLS, tag information is added to the RESV message. In this way, when RSVP reserves resources in the network, it also establishes LSP that meets QoS requirements. Taking Hierarchical Mobile MPLS as an example, when a communication node sends a PATH Message to MH, The PATH Message is forwarded to the FDA, and the FDA then forwards the PATH Message to the FA in the subnet of the MH. FA assigns a tag as its inbound tag based on the information in the PATH, and then sends a RESV message containing the tag value to the FDA. After receiving the RESV message from FA, the FDA uses the tag value and the port on which the RESV is received as the LSP outbound label and outbound port between the FDA and FA, then, a tag is assigned as the incoming tag, and a RESV message with the tag information assigned by the FDA is forwarded to the communication node. When RESV arrives at the communication node, the communication node uses the label value in the message and the port from which the RESV is received as the LSP outgoing label and outgoing port between the communication node and the FDA. So far, the QoS-compliant LSP from the communication node to MH is established. When the MH moves to the new subnet, the FA of The New subnet sends a RESV message to the FDA, and establishes a new LSP between the FA and the FDA that meets the QoS requirements of the business flow through a similar process. Because there is already an LSP between the FDA and the communication node that meets the QoS requirements of the business flow, you only need to modify the outbound label and outbound port of the corresponding LSP on the FDA. In the mobile MPLS with IntServ, the PATH and RESV messages can be transmitted on the pre-established signaling LSP. Mobile MPLS with IntServ features high flexibility and good service quality assurance. However, the system overhead and scalability are high in resource reservation, scheduling, and buffer management.

Iii. Jiyu

Mobile MPLS is a new research area. future research directions include: ① study the new network architecture, without changing or changing the existing network facilities as little as possible, implement the mobile MPLS function on the MPLS network; ② how to implement the VPN service based on the mobile MPLS; ③ how to implement the mobile MPLS Traffic Engineering; ④ how to control the flow and control the message format.

Related Articles]

• Analysis of MPLS technology: converged routers and ATM Switches
  

• Detailed introduction to the concept of MPLS Routing Protocol
  

• How multicast is implemented in an mpls vpn Network