In terms of computer networking, mobility means the ability or freedom to change network points of attachment. There have been previous studies conducted to exploit network mobility. Among them are mobility using network layer and link layer. However, there is also a possibility of providing mobility at the Transport layer. A research on transport layer approach to host mobility was conducted in 2002, which allows mobility decisions to be based on end-to-end channel characteristics as well as local connectivity information.
Multiplexing is the main theory behind this formulation. If we assume that all channels are available all the time with an infinite period, adding a channel will be just changing the period from infinite to a finite value, and deleting a channel will be just setting the period to infinity. On the other hand, multiplexing requires special attention on the reliability algorithms. The key idea behind this approach is the use of multiple overlapping technologies and the complete workflow comprises of three main parts
Link Layer Manager
Transport protocols
Location service
Link Layer Manager (LLM) is responsible for the discovery of the link layer and IP layer configuration. A collection of well-defined transport protocol carries the traffic and the location service allow mobile hosts to be contacted when away from the home network. The task is achieved using multiple link layers for a single data flow. This is done using the channel, which is an end-to-end transport layer connection that encompasses all available link layers and multiplexes the data of a single flow into these links. The sending application just sees one transport layer interface which receives the data from the application and sends it through sub-channels. These sub-channels are basically network layer sockets mapped to different link layers. Although the transport layer doesn’t communicate directly with the link-layer but it has to be link-layer aware. To manage communication between these layers, LLM comes into play, which acts as a database for the event channel and an interface for inter-layer communication. Further, two main protocols can be introduced. One as multimedia multiplexing transport protocol (MMTP) and another as reliable protocol. MMTP acts as a rate based multiplexing protocol designed to carry packets with hard deadlines and supports the transmission of time-sensitive rate based data streams. Furthermore, MMTP creates various channels that multiplexes the data into any available communication sub-channel. MMTP being dynamic has the tendency to change or remove any sub-channel. Based on the estimates of bandwidth delay characteristics, the MMTP decides as to which sub-channel can be used for transmitting current packet. On the other hand, R-MTP or reliable protocol works for transmitting bulk data to mobile systems that have access to multiple link layer technologies. To multiplex data into multiple sub-channels it is necessary to know the sub channel characteristics, particularly the available bandwidth. The regularity in which packets are transmitted in R-MTP also allows the inter arrival time to be used as an aid to differentiate congestion losses from medium losses. R-MTP measures the minimum inter arrival time of each channel and multiplexes packets on the channels according to the resulting periods. R-MTP increases the period of individual sub channel in case of congestion and decrease it if more bandwidth is available.
Advantages –
Having mobility at transport layer is advantageous because it overcomes the problems experienced at link layer and network layer. Link layer solutions are limited to point-to-point connectivity choices between the mobile and a base station whereas network layer solutions are limited to information about the points of attachment of the host to the network. However, mobility at TCP allows decision to be based on end-to-end channel characteristics, as well as local connectivity information, which is network layer independent.
Disadvantages –
One of the problems of using TCP in mobile systems is the slow-down of transmission in the presence of losses. TCP has no way of identifying the cause of the loss because it sends packets in bursts.
REFERENCES
Adapting End Host Congestion Control for Mobility, Wesley M. Eddy Verizon Federal Network Systems, Cleveland, Ohio Yogesh P. Swami Nokia Research Center, Irving, Texas, NASA/CR—2005-213838
C. Lochert, B. Scheuermann, and M. Mauve, “A survey on congestion control for mobile ad hoc networks,” Wireless Communications and Mobile Computing, vol. 7, no. 5, p. 655, 2007
Host-to-Host Congestion Control for TCP, Alexander Afanasyev, Neil Tilley, Peter Reiher, and Leonard Kleinrock, UCLA, IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 12, NO. 3, THIRD QUARTER 2010