Chapter 2. Transport Layer Protocols

Transcription

1 Chapter 2 Transport Layer Protocols

2 Chapter 2 Transport Layer Protocols The Internet has been continuously progressing ever since its inception as ARPANET. It has grown from a network of some hundreds of hosts to millions of hosts supporting an ever-increasing number and variety of applications and services. TCP/IP reference model is widely used in the networking. In the chapter, there is brief discussion of OSI model and transport layer. Research prime area is to focus upon transport layer. The chapter contains description and working mechanism of various transport layer protocols- TCP, UDP, TLS, WTCP, WTLS, WAP, WAP DP, SCTP, RSVP, DCCP, XTP, CSP, RUDP, RRP, etc. 2.1 Introduction of OSI model and Layers The Open Systems Interconnect (OSI) model [84], consists seven layers viz. application, presentation, session, transport, network, data link and physical as per the hierarchy as shown in Fig 2.1 The lowest layer known as the Physical Layer, carries out the transmission and reception of unstructured raw bit stream. The next highest level called Data Link Layer, looks after the untroubled transfer of data-frames from one node to another. Network layer, which is placed above Data Link layer and below Transport layer, performs an important job of decision making regarding the choice of path of data while adhering to network conditions, priority of service and other factors. Layer 5 i.e. Session Layer, allows session establishment between processes running on various stations. Presentation layer, works as a translator for the network, thereby formatting and reformatting the data, to be presented to the application layer, between the sending station and the receiving station. Topmost layer or Application Layer, caters several functions such as remote file access, remote printer access, network management, directory services, electronic messaging, etc. and acts a window to users and application processes to access network services. 32

3 In between session layer and network layer of the OSI model is the Transport layer. It carries out end-to-end communication services, thereby ascertaining the error-free and in-sequence delivery of messages without losses. It performs message segmentationbreaking the messages into smaller units and forwarding it to network layer, message acknowledgment for end-to-end message delivery, message traffic control- to avoid buffering and session multiplexing for combining several messages into one stream and reliability. As a rule, the transport layer has a capacity of restoring large messages but network layer restricts it because of which it has to break the message into small segments/ units. Also, it is to be noted that the transport layer provides various levels of reliable delivery service for the application layer depending on the application layer requirements. Request for Comments (RFC) for Communication Layers, is 1122 [85] and updated by 1349, 4379, 5884, 6093, 6298, 6633, and Figure 2.1: OSI Model 33

4 2.2 Transport Layer Protocols With the growing complexity of Internet and wide-scale usage of computers, several applications are run on to computers at the same time. Computers send and receive several messages on a continuous basis. The sending and receiving computers, each carry out the delivery of messages not only from one computer to the next but also from process on one computer to another. Each computer runs analogous stacks, with data being passed down the sender s stack (i.e. from the sending application) to up the receiver s stack (i.e. to the receiving application). The transport layer manages the process-to-process delivery of the entire message. Hence, for the purpose, transport layer header have included two special kind of addresses viz. a service-point address in the OSI model and port address in the Internet with TCP/IP protocol suite. It runs in end systems: sender side breaks message from application layer into segments and passes to network layer whereas receiver side reassembles segments into messages and passes to application layer. It is to be noted, network layer performs logical communication between hosts, whereas transport layer enhances quality of service, which network layer caters, in order to match the application needs and facilitates logical communication between processes. A transport layer protocol could be connectionless as well as connection-oriented. In connectionless, each segment is considered as independent packets first and then transported to transport layer at destination machine. And in Connection-oriented, each segment is transferred after the establishment of connection between transport layers of source and destination. Transport layer is the heart of whole protocol hierarchy. Its function is to facilitate reliable, cost-effective and congestion control delivery of data from source to destination apart from other functions of multiplexing, addressing and framing Transmission Control Protocol Transmission Control Protocol (TCP) [86], is one of the main protocols, which performs the function of delivering stream of bytes from one program to another or from one computer to another in a reliable manner. It establishes full virtual connection between two points. IP address and TCP port number define the endpoints. Its main work is to control the network congestion, which might usually occur during 34

5 the transfer, that occurs as various packets take various routes to reach the destination. It thus ensures reliable delivery of packets between applications. It makes use of the technique of positive acknowledgement and retransmission i.e. it guarantees the entire received bytes to be sent in the same order without specific change. The technique works in such a manner that the sender while sending packets automatically switches on the timer and keeps a record of each packet send. The receiver on the other hand, sends an acknowledgement after receiving the message. If in case, timer expires before the acknowledgement of message than in such case the timer once again sends the message. To carry out a reliable service, TCP uses numbering system (determines bytes, sequence and acknowledgement numbers) as well as control mechanisms (flow, error and congestion controls). It is the main protocol of Internet Protocol suite, complementing the Internet protocol and thus the suite is commonly known as TCP/IP suite. TCP and IP provide the basic structure of Internet. In a network, IP facilitates packet delivery without considering reliable packet delivery. On the other hand, TCP promotes reliable packet delivery, thereby compensating for the deficiency of IP protocol. So, in general whichever application desires to send chunk of data it simply issues request to TCP and TCP handles the IP details. The Transmission Control Protocol is used in applications such as World Wide Web, , file administration, etc. Transmission control protocol is described in RFC User Datagram Protocol David P. Reed designed User Datagram protocol (UDP) [87] in It is one of the main protocol in Internet Protocol suite. It serves the application layer and network layer like TCP protocol i.e. it transfers messages usually known as datagrams from one application having port number to another. However, there is no guarantee of the message being sent to reach the destination. It might get lost or corrupted or duplicated on the way, as it is incapable of establishing a proper connection as TCP does and hence it is also referred as connectionless protocol because of less reliability. Also it doesn t employ ACK, flow and congestion control like TCP and hence is one of the reason for no guarantee of message reaching the destination in the same 35

6 sequence or order as transmitted. UDP is comparatively simple, in the sense. In TCP, data is send usually in a stream with no distinction between two packets as they move in a stream. However in UDP, packets are sent individually i.e. one packet per one read call. The only overhead, which UDP adds to the packet is to establish process to process communication, rather than host to host communication of IP layer. Therefore, UDP is very suitable for small message communications or applications, which do not need strong reliability. These are mainly client/ server request/ reply or video conferencing. Some of the known applications of UDP are streaming media, real-time multiplayer games and voice over IP. Also voice and video-trafficking is sent usually using UDP. User Datagram Protocol is described in RFC Transport Layer Security Transport Layer Security (TLS) [88] is a protocol, which helps to secure data as it provides communication security over Internet i.e. it ensures privacy between communicating applications (when server and client communicates, it makes sure no other party is hampering the flow of communication). Secure Sockets layer (SSL) is the predecessor of TLS and the difference between them is so minor, hence usually they are lumped together as TLS/SSL. The Transport Layer Security is composed of two layers viz. TLS Record Protocol and TLS Handshake Protocol. TLS Record Protocol is used for the purpose of ensuring security with the use of Data Encryption Standard. On the other hand, Handshake Protocol provides encryption algorithm and cryptographic keys to client and server for the facilitation of secured communication between them, thus there is no forgery of message or no eavesdrop by the third party. Although no single security measure could secure the data from being getting hampered but implementing security protocols like TLS could reduce the chance of such threats. TLS has a wide range of applications, which includes its implementation in HTTP, FTP, SMTP, NNTP and XMPP. Also, it is being implemented with Transport Layer Protocol (TCP), User Datagram Protocol (UDP) and Datagram Congestion Control Protocol (DCCP). Electronic commerce and asset management also makes an application of TLS. It is widely used in various open source software projects. Secure 36

7 Sockets Layer (SSL) [89] Version 2.0 is described in RFC 6176 with updates in RFC 2246, 4346, 5246 and the TLS Protocol Version 1.0 is described in RFC Stream Control Transmission Protocol Stream Control Transmission Protocol (SCTP) [90], a transport layer protocol, is an improvement of TCP and UDP. It uses the key features of both TCP (i.e. reliability, sequential transmission of message with congestion control) and UDP (i.e. message oriented). However, SCTP has two additional features: Multi-homing and Multistreaming. These two features of SCTP could be explained as follows: Multi-homing enables system, which has multiple interfaces to avoid time-delays, which usually occur in telecommunication network. It is also used to carry out transmission of message. On the other hand, Multiple-streaming in SCTP could be understood as the facilitation of multiple streams through connection on contrary to TCP, where one single data stream is allowed per connection. SCTP s multi-streaming is also used with infrastructures, which connect multiple means of communications simultaneously. SCTP facilitates the operation of message transfer by keeping message and control information into separate chunks (i.e. data chunk and control chunk), where each chunk carries data of only one user, identified by chunk header. These chunks are bundled into packets. Stream Control Transmission Protocol is described in RFC 4960, while obsoletes RFC are 2960, 3309 and updated RFC are 6096, 6335, Resource Reservation Protocol Resource Reservation Protocol (RSVP) [91], is a transport layer protocol also a part of Internet Integrated Service (IIS) model. It is designed to reserve resources across network. It primarily lays out principles, to direct channels and the paths on the Internet to be secured for multicast and bandwidth messages. RSVP ensures besteffort service, real-time service and controlled link-sharing. It is similar to vivid control protocols in some manner viz. ICMP/IGMP. It is described in RFC

8 Either hosts use it for requesting Quality of Service from network or routers use it to deliver Quality of Service requests along the flows. RSVP services generally result in resources being reserved in each node along the path. The twin key concepts of RSVP model are flowspec (QoS specific information is called a flowspec) and filterspec ( defines the set of packets to be affected by a flowspec). RSVP also comprises two messages of which the first is path message, which is sent from sender host along the data path and stores the path state in each node along the path. The second one is reserved message, which is sent from the receiver to the sender host along the reserve data path. Resource Reservation Protocol (RSVP) is described in RFC 2205 and updated by RFC 2750, 3936, 4495, 5946, 6437, Datagram Congestion Control Protocol Datagram Congestion Control Protocol (DCCP) [92], is a message-oriented transport layer protocol. It serves the purpose of establishing, maintaining as well as tearingdown of an unreliable packet flow and congestion control of the packet flow. It is suitable for applications, which carry out the transfer of large amount of data as well as those, which benefits from control over trade-off between timeliness and reliability. DCCP is useful for applications with time constraints on the delivery of data. Such applications include streaming media, multiplayer online games and Internet telephony. It implements reliable connection setup, tear-down, Explicit Congestion Notification (ECN), congestion control and feature negotiation. DCCP provides a way to gain access to congestion control mechanisms without having to implement them at the application layer. It allows flow-based semantics like in Transmission Control Protocol (TCP), but does not provide reliable in-order delivery. RFC 4340 describes Datagram Congestion Control Protocol (DCCP) and 5595, 5596, 6335, 6773 updates it. 38

9 2.2.7 Multipath TCP Multipath TCP (MPTCP) [93], is an on-going effort of the IETF's Multipath TCP working group, which allows a TCP connection to use multiple paths to maximize resource usage and increase redundancy. TCP/IP communication is currently restricted to a single path per connection, yet multiple paths often exist between peers. The simultaneous use of these multiple paths for a TCP/IP session would improve resource usage within the network and, thus, improve user experience through higher throughput and improved resilience to network failure. Multipath TCP provides the ability to simultaneously use multiple paths between peers. The protocol offers the same kind of service to applications as TCP (i.e., reliable byte stream), and it provides the components necessary to establish and use multiple TCP flows across potentially disjoint paths. Multipath TCP (MPTCP) is a set of extensions to regular TCP to provide a Multipath TCP [2] service, which enables a transport connection to operate across multiple paths simultaneously. In January 2013, the IETF published the Multipath specification as an Experimental standard in RFC Licklider Transmission Protocol Licklider Transmission Protocol (LTP) [94], is a point to point protocol for the use in deep space links. LTP is designed to run directly over a data link layer (such as e.g. AOS), but it could also run on the User Datagram Protocol (UDP). LTP is commonly seen as the standard underlying convergence layer protocol for the Bundle Protocol (RFC 5050), supporting a wide selection of networks. The Licklider Transmission Protocol (LTP), designed to provide retransmission-based reliability over links, is characterized by extremely long message round-trip times (RTTs) and/or frequent interruptions in connectivity. Communication across interplanetary space is the widely prominent example of such sort of environment, 39

10 LTP is principally aimed at supporting "long- haul" reliable transmission in interplanetary space, but it has applications in other environments as well. Since, no mechanisms for flow control or congestion control are included in the design of LTP, the protocol is not intended or appropriate for ubiquitous deployment in the global Internet. Licklider Transmission Protocol is described in RFC 5325 and Wireless Transmission Control Protocol Wireless Transmission Control Protocol (WTCP) is a proxy based modification of TCP, which is being used in wireless networks to improve the performance of TCP. It is being created to provide an easy and standard means for transmitting messages (alphanumeric and binary data) to and from the wireless network gateway over a wired network. In other words, it transfers data, both one-way and two-way, to appropriate receiving devices. WTCP is usually placed in an intermediate gateway between source host and mobile host in such a manner that the hosts aren t even aware about the existence of WTCP. The working of WTCP between two hosts is explained in the following manner; when any mobile hosts send message through TCP, it is received by WTCP at the base station and it forwards it over the network to the awaiting source host. On the reception of message, the source hosts sends an acknowledgement back to the base station host through the network, which is then forwarded to the mobile host. So, in general WTCP exactly performs the same role as TCP does, but instead of replacing it completely, WTCP works with TCP in order to enhance the performance by handling high bit error rates. Hence, it is to be inferred, WTCP doesn t replaces TCP on the hosts. On the contrary, it is placed on a proxy between two communicating hosts. It significantly improves the throughput because of its unique features. So, in order to increase the out-turn of data transmission, Wireless Transmission Control Protocol (WTCP) is used as it is capable of handling vulnerable problems occurring over wireless networks. It has the feature of fast acknowledgement, ondemand retransmission and time-out mechanisms. However, WTCP is fairly new protocol and it is not widely yet accepted. 40

11 Wireless Application Protocol According to Wireless Application Protocol (WAP) Forum, it is a worldwide standard for providing Internet communications and advanced telephony services on digital mobile phones, pagers, personal digital assistants and wireless terminals. It is the result of the joint effort of companies like Ericsson, Motorola, Nokia and Unwired Planet (commonly known as WAP group, which created a forum later on whose membership is opened to all) to create a standard for facilitating advanced services within wireless network. WAP is composed of four layers viz. Wireless Application Environment (WAE), Wireless Session Layer (WSL), Wireless Transport Layer Security (WTLS) and Wireless Transport Layer (WTP). Prior to the introduction of WAP, the mobile service providers had greater difficulty to provide interactive data services such as by mobile phones, trafficking of stockmarket prices, sports results, news headlines, music downloads, etc. In fact one could say, internet was Internet (i.e. Only computers used to carry out the net surfing work) and mobile phone was mobile phone (i.e. no application of Internet facility). But the problem has been counteracted with the emergence of WAP, which enabled availability of massive information, communication and data resources of the Internet on mobile phones. It is also referred to as a technical standard for accessing information over a mobile network. It is to be noted, just like a standard internet site is used for browsing purpose, to access WAP enable website, one need a micro browser called WAP Browser for mobile devices and mobile networks WAP Datagram Protocol Wireless Datagram Protocol (WDP), a protocol in WAP architecture, defines transmission layer protocol for an internet model and enables WAP to be bearerindependent. In other words, it adapts the transport layer of the underlying bearer service, it offers to the upper layer and invisible interface independent of the underlying network technology used, and presents consisting data format to the higher layer of the WAP protocol stack. As WDP allows only its transport layer to deal with physical network-dependent issues, using mediating gateways it could acquire global 41

12 interoperability. It resembles to UDP and carries out movement of information from receiver to the sender Wireless Transport Layer Security Wireless Transport Layer Security (WTLS) is a security protocol and is derived from TLS (successor of SSL), with an aim to address problematic issues such as limited processing power, memory capacity, low bandwidth in the wireless network. It makes use of modern cryptographic algorithms and allows negotiations of cryptographic suites between client and server just like TLS, thereby ensuring adequate authentication, data integrity and privacy protection mechanisms between applications communicating using Wireless Application Protocol (WAP). Many organisations such as defence, banks or health organisations, etc. require high level of security for its data. In order to carry out wireless transmission of data between these organisations and their destination, such organisations demand highly secured encryption so as to protect communication form attack or forgery during transmission. WTLS is designed to support wireless networks. It is done through facilitation of secured messages in high latency and low bandwidth environment. WTLS has both advantage as well as disadvantage. On the one hand, it offers the necessary support for the limited memory and processing capabilities of WAP-enabled devices, which are failed to be address by SSL and TLS; and on the other hand, it sometimes allows weak encryption algorithm, thereby harnessing the security of user. Inspite of the security problems, it is considered to be correct security solution. So, with a little development and certain improvement, WTLS intends to provide a correct security solution Multipurpose Transaction Protocol Multipurpose Transaction Protocol software is a proprietary transport protocol (OSI Layer 4) developed and marketed by Data Expedition, Inc. (DEI). DEI claims that MTP offers superior performance and reliability when compared to the Transmission Control Protocol (TCP) transport protocol. Taking cues from the high-volume, 42

13 request-response pattern of modern network applications, Core MTP follows a simplified transaction data model. Each core network operation consists of small request datagrams exchanged for a potentially huge collection of response datagrams. Combined with a more robust packet design, the aforesaid model eliminates overhead like three-way handshakes and time-wait states. These core transactions might then be modularly combined to create more sophisticated data models, with only the minimum overhead appropriate to the task. For many MTP users, thus it translates into seven times faster data transfers over high-speed WANs, and ten times improved reliability on no matter which network, when compared to even modern TCP implementations. Thus, it holds viable even for data, which is already compressed and completely unique, where compression and caching would break down. Also importantly, the results are scalable; the adaptive nature of the protocol means no setup tuning or special equipment is needed Xpress Transport Protocol (XTP) Xpress Transport Protocol (XTP) is a transport layer protocol for high-speed networks. XTP Forum promotes the protocol and developed it to replace TCP. XTP is a reliable, real-time, lightweight protocol. XTP provides protocol options for error control, flow control, and rate control. Instead of separate protocols for each kind of communication, XTP controls packet exchange patterns to produce various models, e.g. reliable datagrams, transactions, unreliable streams, and reliable multicast connections. Long latency is one of the major problems in satellite communications. Coupling it with possible environmental variables and sometimes asymmetrical bandwidth conditions, the quality of service in satellite communications is sometimes lacking. XTP addresses these issues in a variety of ways such as a Selective Retransmission algorithm, which deals with loss recovery. XTP does not employ congestion avoidance algorithms. Current transport layer protocols, such as DoD's Transmission Control Protocol (TCP) and ISO's Transport Protocol (TP) are not being designed for the next generation of high speed, interconnected reliable networks viz. FDDI and the gigabit/second wide 43

14 area networks. Unlike the other previous transport layer protocols, XTP is designed to be implemented in hardware as a VLSI chip set. By streamlining the protocol, combining the transport and network layers and utilizing the increased speed and parallelization possible with a VLSI implementation, XTP is able to provide the endto-end data transmission rates demanded in high speed networks without compromising reliability and functionality. XTP provides the reliable transmission of data in an inter-networked environment, with real-time processing of the XTP protocol i.e., the processing time for incoming or outgoing packets is no greater than the transmission time. XTP contains error, flow and rate control mechanisms similar to those found in other more modern transport layer protocols in addition to multicast capability. Timer management is minimized in XTP there is only one timer at the receiver, used in closing the context. XTP has a 32 bit flow window. XTP's state machine is specifically designed for parallel execution. Address translation, context creation, flow control, error control, rate control and host system interfacing can execute in parallel Cubesat Space Protocol (CSP) Cubesat Space Protocol (CSP) is a small network-layer delivery protocol designed for Cubesats. A group of students from Aalborg University in 2008 developed the idea, and further developed for the AAUSAT3 Cubesat mission, successfully launched in The protocol is a 32-bit header containing both network and transport layer information. Its implementation is designed for, but not limited to, embedded systems such as the 8-bit AVR microprocessor and the 32-bit ARM and AVR from Atmel. The implementation is written in C and is currently ported to run on FreeRTOS and POSIX and pthreads-based operating systems such as Linux. Support for Mac OS X and Microsoft Windows will be available in version 1.1. It supports transparent forwarding of packets over e.g. space link. CSP supports for both connectionless operation (similar to UDP), and connection oriented operation (RFC 908 and 1151). 44

15 Reliable User Datagram Protocol (RUDP) In computer networking, the Reliable User Datagram Protocol (RUDP) is a packet based transport layer protocol designed at Bell Labs for the Plan 9 operating system. It aims to provide a solution, where UDP is too primitive because guaranteed-order packet delivery is desirable, but TCP adds too much complexity/overhead. In order for to gain higher Quality of Service, RUDP is implemented, which is similar to TCP with less overhead. RUDP is rested upon RFCs 1151 and Reliable Data Protocol. RUDP is layered on the UDP/IP Protocols and provides reliable in-order delivery (up to a maximum count of retransmissions) for virtual connections. A reliable transport protocol is needed to transport the telephony signalling across IP networks. The reliable transport is able to provide architecture for a variety of applications (i.e. signalling protocols) requiring transport over IP. Existing IP protocols have been scrutinized and it has been concluded that RUDP is designed to allow characteristics of each connection to be individually configured, thus many protocols with independent transport requirements could be implemented simultaneously on the same platform. RUDP has a very flexible design, which makes it suitable for a variety of transport uses TCP Westwood TCP Westwood (TCPW) is a sender-side-only modification to TCP New Reno, which is intended to better handle the large bandwidth-delay product paths (large pipes), with potential packet loss because of transmission or other errors (leaky pipes), and with dynamic load (dynamic pipes). TCPW fully complies with the end-to-end TCP design principle. The key innovative idea is to continuously measure, at the TCP sender side, the bandwidth used by the connection via monitoring the rate of returning ACKs. The estimate is then used to compute congestion window and slow start threshold after a congestion episode, i.e. after three duplicate acknowledgements or after a time-out. The rationale of such a strategy is plain: In contrast with TCP Reno, which blindly halves the congestion 45

16 window after three duplicate ACKs, TCP Westwood attempts to select a slow start threshold and a congestion window, which are consistent with the effective bandwidth used at the time congestion is experienced. The proposed mechanism is particularly effective over wireless links, where sporadic losses out of radio channel problems are often misinterpreted as a symptom of congestion by current TCP schemes and thus, lead to an unnecessary window reduction TCP Vegas TCP Vegas is a TCP congestion avoidance algorithm, which emphasizes packet delay, rather than packet loss, as a signal to help to determine the rate at which to send packets. TCP Vegas detects congestion at an incipient stage resting on increasing Round-Trip Time (RTT) values of the packets in the connection. The algorithm depends heavily on accurate calculation of the Base RTT value. TCP Vegas is one of a series of efforts at TCP tuning, which adapt congestion control and system behaviours to new challenges faced by an increase in available bandwidth in Internet components on networks like Internet. There are five techniques, which Vegas employs to increase throughput and decrease losses. The first two of these techniques are mere extensions, which result in a more timely decision to retransmit a dropped segment. The third technique is an adjustment to TCP s congestion window resizing algorithm; it is required because of Vegas more accurate time-out mechanism. The fourth technique is designed to better space transmissions. The fifth technique is widely novel. It gives TCP the ability to anticipate congestion, and adjust its transmission rate accordingly The Internet Link protocol The Internet Link protocol or IL is a connection-based transport layer protocol, designed at Bell Labs originally as part of the Plan 9 operating system. It is assigned the Internet Protocol number 40. It is similar to TCP but much simpler. Its main features are reliable datagram service, in-sequence delivery, Internetworking using IP, low complexity, high performance, and adaptive time-outs. 46

17 IL is a lightweight protocol encapsulated by IP. It is connection-based and provides reliable transmission of sequenced messages. No provision is made for flow control since the protocol is designed to transport RPC messages between client and server, a structure with inherent flow limitations. A small window for outstanding messages prevents too many incoming messages from being buffered; messages outside the window are discarded and hence to be retransmitted. Connection set-up uses a twoway handshake to generate initial sequence numbers at each end of the connection; subsequent data messages increment the sequence numbers to allow the receiver to resequence out of order messages. In contrast to other protocols, IL avoids blind retransmission. Thus, it helps performance in congested networks, where blind retransmission could cause further congestion. Like TCP, IL has adaptive time-outs, so the protocol performs well both on the Internet and on local Ethernets. A round-trip timer is used to calculate acknowledge and retransmission times, which matches the network speed. 47

18 Chapter 3 Security and Priority issues of Quality of Service at Transport Layer