Performance of Mobile IPv6 fast switching in WLAN (1)

When a mobile node switches over a zone, the wireless link must be switched first. If the new link and the new link are not in the same IP subnet, IP subnet switching is also required. Even if the routing optimization technology is adopted, the packet delay during wireless link Switching and subnet switching is considerable, the latency is mainly caused by end-to-end Mobile IP address registration after link switching. During the switchover, the Group sent to the mobile node may be lost. Therefore, the quick switchover solution will help improve the service quality of the group data. The link switching is usually completed by the Layer 2 protocol or hardware switching, and the switching over the IP subnet must be completed by the Layer 3 protocol or soft switching.

Currently, there are many documents on fast switching of Mobile IP addresses. The basic ideas include three types of multicast, pre-Switching Based on Layer 2 mobile triggering, and grouping cache, this article mainly discusses the concept of pre-Switching Based on the second layer of mobile triggering, that is, before a mobile node moves to a new link, it first initiates a switching procedure to obtain the forwarding address on the new link in advance. The switching procedure is achieved by exchanging new messages between the new and old access routers and between the Access Router and the mobile node. This method requires the mobile node to know in advance that it is about to move to the new network, so the second layer of support is required. We know that AP in WLAN knows whether to perform L3 switching from beacon or inquiry response before L2 switching, and mobile IP registration is time-consuming, therefore, we need to find a way to save time in a link and resume communication earlier. The main idea of this mobile IPv6 fast switching is to use the link layer information for L3 switching, and to get an IP connection in advance by passing data between the new and old routers. The principle of its application is: data is transmitted at the receiver layer before the network layer. Therefore, the link layer obtains information earlier than the network layer.

1. System Modeling

1.1 Quick Switch procedure

First, describe the steps for fast switching of the draft:

(1) MN needs to perform L3 switching from the received ESSID, so it sends a proxy routing request to the original AP, including the link layer address of the new AP.

(2) The original AR maps the link layer address of the new AP to the IP address of the new AP, which requires a ing table at the original AR.

(3) The original AR determines whether MN uses the passive (stateful) or the active (stateless) address in the new AP.

Active address: the original AR uses the MN Interface ID on the New subnet to obtain a new COA (careofaddress). Generally, this interface ID is the 48-bit MAC address of MN, the COA is formed by adding the prefix of the New subnet, and sent to MN through HI/HACK. At the same time, you must first check whether the address is unique. The "unique address confirmation request" is sent at the same time as HI/HACK. If it is not unique, the passive address is used.

Passive address: the original AR uses the HI/HACK exchange to send a new AR request address. In this case, the new AR usually uses DHCP to allocate the address.

(4) The original AR returns a proxy routing message to MN. This message is sent simultaneously with HI/HACK when the stateless Address is configured, and then after HI/HACK when the stateful address is configured.

(5) MN then sends a fast Binding Update (F-BU) to the original AR and binds it to the new COA. Note that this is L3 information, which is performed before L2 switching.

(6) now MN starts L2 switching. By the end of L2 switching, MN has been moved to the new AP.

(7) MN will receive the F-BACK, indicating that the original AR received his F-BU, then you can use the new COA communication, the information is sent at the same time on the new link, that is, the original AR-new AR-New AP-MN; the old link, that is, the original AR-original AP-MN ). In this way, the packets from the original AR will be sent to the new AR according to FBU.

Note that step 15 of L2 switching is completed by hardware, so that MN can send IP layer information such as F-BU can not start L2 switching, that is, the interface hardware cannot start switching without the permission of MN. When the entire switchover is completed, we can see that this is more effective than using mobile IP registration, saving the switching time.

1.2 Algorithm Improvement

It can be noted from the proposed switching algorithm that both L2 switching and L3 switching are controlled by MN, so in order to ensure normal communication between MN, A new COA must be obtained before L2 switchover, and then wait for the L2 primary switchover steps to be completed before binding and updating. In this way, L2 and L3 switchover can be performed in less steps at the same time. However, if L3 switching is controlled by the network, the situation is different:

(1) When the switching condition is met, MN sends a query request, and all the APs that receive the query request respond to one query response. MN selects the appropriate new access AP based on the information in the response.

(2) MN sends a switching trigger to both APs at the same time, announcing the formal start of L2 switching. Considering that MN cannot communicate before it connects to the new AP, if the network does not have enough information, it cannot determine whether to perform the L3 switching. Therefore, the switching trigger should contain the MAC addresses of the New and Old APs. The original AP receives the switching trigger, pass this information to the original AR immediately. When a new AP is triggered, it starts the L2 Switch mentioned above and starts to synchronize with MN through information exchange.

(3) when the original AR receives L2 switching, it must be able to identify the subnet of another AP. This requires that the MAC address set of the AP contained in each subnet be retained at the AR, in addition, the MAC addresses of different subnet AP are different.

(4) If the original AR finds that the two APS belong to different subnets (two APs are connected to different routers), the request address of the new AP is the AR (new AR, the new AR allocates a COA address based on the Dynamic Host Configuration Protocol and returns it to the original AR. The original AR sends the address from two bss to the mn to ensure that it can receive the address. This is because: MN may be connected to the new AP (Link Layer switching is completed), but it may still be connected to the original AP. At the same time, the link layer is switched. As soon as MN enters the new network, the COA address is available. It should be noted that the original AR actually acts as a home agent.

(5) afterwards, MN sends the Binding Update to the original AR (hometown proxy) for registration, and then receives the binding response. That is, the registration is successful.

Note that the new AR is not a foreign proxy in this switching scheme, because if it is a foreign proxy, it must perform a series of validity checks on the registration request when registering the request. If one of the checks fails, the foreign proxy sends a registration Response Message to the mobile node to reject the registration request. If the check succeeds, the foreign proxy completely removes the IP header and UDP header of the packet containing the registration request message, and adds the new header before sending it to the hometown proxy. Similarly, when a foreign agent receives a registration response from a home agent, a series of validity checks will be performed on the message. Once the response is invalid, a registration response containing the appropriate Code domain is generated and sent to the mobile node. Here, the new AR is only used as a router.

From the above steps, we can see that, on the one hand, we use the link layer information to initiate a switchover (the idea of the draft), saving some time. On the other hand, because the COA address is obtained and the L2 address is switched at the same time, it saves much time, especially for WLAN. Because all its control signals and data signals share one channel, and the MAC layer adopts the CSMA/CA policy, if several MN instances in the same BSS work simultaneously, L2 switching takes a long time. From the simulation results, when there are 4 MN instances, the L2 switching time is about 2 s. Using these 2 s time to obtain COA will greatly reduce the switching latency.