Karpagam Journal of Computer Science, Volume 07, Issue 03, March April 2013, pg. no. 168 to 183, ISSN 0976-2926
Experimental Performance Testing of TCP and UDP Protocol over WLAN Standards, 802.11b and 802.11g. Dr. Atul M Gosai#1, Bhargavi H Goswami*2 #
Department of Computer Science, Saurashtra University, Rajkot, Gujarat, India * Department of Computer Science, NIMS University, Jaipur, Rajasthan, India
1
[email protected],
[email protected]
Abstract— Wireless local area networks (WLANs) based on the IEEE 802.11 standards, also known as WiFi, are popular in enterprise networks, homes, and public hotspots such as airports and hotels. WLAN enables wireless networks that support data rates of 1–54 Mb/s over small areas of a few thousand square meters are widely deployed all over world. In this article we discuss WLAN architectures for providing internet service capability across widely deployed 802.11based networks. Specifically, we present two design choices for service: 802.11b and 802.11g, and recommend the latter as a preferred option. We have implemented these two WLAN standards over two widely used protocols TCP and UDP covering the applications based on connection oriented as well as connection less and reliable and non reliable service providing protocols. For comparative study, we have implemented our protocol over NS2 simulator and we have presented our experimental results in the form of graphs for the purpose of analysis of performance parameters. Keywords— WLAN, 802.11b, 802.11g, TCP, UDP, Performance Parameters like Throughput, Queue Length, End to End Delay, Drop Rate.
1. INTRODUCTION Wireless networks offer wide options of utility by both home and corporate users. Before, wireless networks were implemented only where the wired networks are not possible or where the wireless network is the only option available considering the security for implementation [4]. But now, wireless gadgets are so much in demand, which is clearing inspiring wide usage of wireless networks. Again, it is a method by which homes, telecommunications networks and enterprise (business) installations can stay away from the costly process of introducing cables into a structure, or as a connection among various equipment locations [4]. There are many types of wireless networks including WPAN, Wireless Mesh Networks, WMAN and WLAN. In this paper we bind our research to WLAN only. Wireless local area networks (WLANs) is based on the IEEE 802.11 standards, also known as WiFi, are popular in enterprise networks, homes, and public hotspots such as airports, college campus, government departments and hotels. WLAN enables wireless networks that support data rates of 1–54 Mb/s over small areas of a few thousand square meters [1]. The reputation of wireless local area networks (WLANs) has led to extensive deployments across different establishments and communication infrastructure. Large-scale deployments are includes large statistics of wireless termination points (WTPs) and it is covering ample areas worldwide and these are increasingly widespread [3]. Wireless network testing for large network is
difficult. With this reason, most researchers use simulation tools. Again developing a good simulation model is a lot of work. Here we figure out how to model the network topology, demonstrate the model in concern of research reference we are using, then we develop topology and verify its parameters, set the objectives of observation and verify its validity, then data collection is done for corresponding real system to set the simulation model's input parameters, simulation model is verified, that it is as expressed in the software, is working perfectly, and then validate the simulation's output adjacent to the corresponding output from the real method of experimentation. After all this, we can say that configuration of modelling is done in a better way [2]. Having given considerations, the focus of this paper is to evaluate the IEEE 802.11b and 802.11g performance using simulation with TCP and UDP protocol with respect to few of the performance parameters like Response Time, Queue Length, Throughput, etc. Rest of the paper from section 2 to 5 describes this.
2. IEEE 802.11 WLAN STANDARDS: The set of standards IEEE 802.11 is for implementing wireless local area network (WLAN) computer communication in the 2.4, 3.6 and 5 GHz frequency bands [5]. Out of all, widely accepted were defined by the 802.11b and 802.11g protocols. We know that these protocols were amendments to the original standard. According to survey 802.11-1997 was the first wireless networking standard, but 802.11b was the first widely accepted one and then followed by 802.11g and 802.11n. There after came 802.11n which is a new multistreaming modulation technique. Other standards in the family (c–f, h, j) are service amendments and extensions or corrections to the previous specifications [6]. We all know that 802.11b and 802.11g use the 2.4 GHz ISM band. We need to note that because of this choice of frequency band, 802.11b and 802.11g equipments at times may occasionally suffer intervention from cordless telephony, microwave ovens and Bluetooth devices. Now as the solution, 802.11b and 802.11g control its vulnerability and interference by using direct-sequence spread spectrum (DSSS) and orthogonal frequency-division multiplexing (OFDM)[14] signalling methods, respectively [5]. Out of available protocols of IEEE 802.11 standards, we concentrate upon 802.11b and 802.11g standards. Describing briefly the standard of 802.11b and 802.11g:
Karpagam Journal of Computer Science, Volume 07, Issue 03, March April 2013, pg. no. 168 to 183, ISSN 0976-2926 A. 802.11b: It was released in 1999 and working on frequency band 2.4. The remarkable increase in throughput of 802.11b along with instantaneous considerable price reductions led to the speedy recognition and acceptability of 802.11b as the ultimate and definitive wireless LAN technology. 802.11b control its interference and defencelessness to interference by using signalling method of direct-sequence spread spectrum (DSSS). It uses the same CSMA/CA media access method defined in the novel standard of IEEE. Due to the CSMA/CA protocol overhead, in practice the maximum 802.11b throughput that an application can achieve is about 5.9 Mbit/s using TCP and 7.1 Mbit/s using UDP [7][8][9]. Devices operating in the 2.4 GHz range include: microwave ovens, Bluetooth devices, baby monitors, and cordless telephones [7]. However they suffer from limited coverage range of AP, resulting in frequent handoffs, even in moderate mobility scenarios[10][11]. 802.11b devices also suffer interference from other products operating in the 2.4 GHz band [6].
B. 802.11g: It was released in 2003 and it is working on frequency band 2.4 same as 802.11b. 802.11g is an improvement to the IEEE 802.11 specification with additional support that extended throughput to up to 54 Mbit/s. The modulation scheme used in 802.11g is orthogonal frequency-division multiplexing (OFDM) [6][14]. Even though 802.11g operates in the same frequency band as 802.11b, it can achieve higher data rates because of its heritage to 802.11a and OFDM [12][14]. Due to the aspiration for higher velocities of standards and reductions in manufacturing expenses 802.11g was widely accepted, time honoured which later slowly steadily replaced 802.11a/b. Despite its major acceptance and popularity, 802.11g still suffers from the same interference as 802.11b in the already crowded 2.4 GHz range [9]. Devices operating in this range include Bluetooth devices, microwave ovens, digital cordless telephones and baby monitors which can show the way to interference issues[13]. In addition to this the sensational achievement of this standard has resulted to bigger usage and compactness problems related to crowding in highly populated urban areas. In spite of the wide range of applications of IEEE 802.11, it is necessary to be aware of some performance problems that may appear in large infrastructure WLANs composed by several Access Points (AP) and several performance criteria into the considerations specified in the next section.[13].
3. NETWORK PARAMETERS: Wireless network performance and its functionality depend on network parameters. The main parameters are: a) b) c) d) e)
TCP UDP Queue Type Throughput Bandwidth
f) Dropped Packets g) Routing Protocols h) Response Time a) TCP: The Transmission Control Protocol (TCP) is one of the foundation protocols of the Internet Protocol Suite. Connection Oriented, Reliable Communication are the characteristics of first type called TCP [15]. TCP provides reliable, ordered delivery of a byte streams from an application on one computer to another application on another computer. TCP is the protocol used by foremost Internet applications such as the remote administration, World Wide Web, email and file transfer [16]. b) UDP: Another type of communication is called UDP which is characterised by Connectionless Unreliable communication. The User Datagram Protocol (UDP) is one of the foundation protocol members of the Transmission Control Protocol & Internet Protocol (TCP/IP) Suite other then set of network protocols used for the Internet [17]. With UDP, computer applications can send messages. Packets are referred to as datagram in case of UDP. UDP sends Datagrams to other hosts on an Internet Protocol (IP) network without the need of earlier communications to set up special transmission channels or data paths. UDP uses a simple transmission model without understood handshaking dialogues without executing for providing reliability, ordering and data integrity [18]. Thus, UDP provides an unreliable service and datagram may arrive out of order, appear duplicated, or go missing without notice[19]. Time-sensitive applications often use UDP because when packet delivery time is the interest, dropping packets is preferable instead of waiting for delayed packets. i.e. Multimedia data transmission, News websites, etc. [17]. c) Queue Type: Defining queue in computer networking, is a group of data packets together waiting to be transmitted by a network device using a per-defined arrangement methodology. Queuing theory is a tool for studying several performance parameters of computer networks and is principally functional in locating the basic causes behind consequences like ―bottlenecks,‖ compromised computer networks performance because of hefty data waiting to be transmitted on at a specific determined criteria phase. Queue size and waiting time can be looked at, or items within queues can be studied and manipulated according to factors such as priority, size, or time of arrival. Queue type/object is a general class of objects capable of holding and marking or discarding packets as they travel through simulated topology [20]. Some of the configuration parameters used for queue objects are: Limit: Limit is the queue size in packets, generally 50. Blocked: It is set to false by default; this is true if the Queue is blocked. Unblocked on Resume: Set to true by default, indicates queue should unblock itself as the time last packet sent has been transmitted.
Karpagam Journal of Computer Science, Volume 07, Issue 03, March April 2013, pg. no. 168 to 183, ISSN 0976-2926 Other queue objects defined by ns group of ISI, derived from base class Queue are Drop-Tail, FQ, SFQ, DRR, RED CBQ and CBQ/WRR objects[21]. 1.
2. 3. 4. 5.
6.
7.
Drop Tail: Drop-Tail objects implement simple FIFO queue. No more methods, configuration parameters or state variables that are specific to drop tail objects. FQ Objects: FQ objects implement Fair Queuing. There are no methods that are specific to FQ objects. SFQ Objects: SFQ objects are a subclass of Queue objects that implement Stochastic Fair Queuing. DRR Objects: DRR objects implement Deficit Round Robin scheduling. RED Objects: RED Objects implement Random Early Detection gateways. The object can be configured to either drop or ‗mark‘ packets. There are no methods that are specific to RED objects. CBQ Objects: CBQ Objects implement Class Based Queuing. They implement packet by packet round robin scheduling among classes of same priority level. CBQ/WRR Objects: CBQ/WRR Objects are a subclass of CBQ objects that implement Weighted Round Robin scheduling among classes of same priority level [21].
Out of the all stated above queue objects, our paper is tested upon two of them. First one was Drop-Tail and second was RED (Random Early Detection). Both have minor difference in performance. After analysing the obtained statistics we can conclude that the performance is better with RED and in comparison of RED performance degrades with Drop-Tail as simulation time passes by [21]. d) Throughput: Throughput in terms of computer networks can be given as average rate of successful packet delivery over a communication channel. Throughput is nothing but actually received speed out of maximum bandwidth available. This term also keep up a correspondence to digital bandwidth consumption [24]. Performance testing can verify that a system meets the specification claimed by our simulation and topology. These performance testing can be in terms of bandwidth, delay, data transfer rate, throughput, efficiency or reliability [22][23]. e) Bandwidth: Bandwidth is a rate of data transfer, bit rate or throughput, measured in bits per second (bps). Bandwidth is the maximum available channel bit rate for our network. Bandwidth purely depends on WLAN 802.11 standard. It is necessary to determine the maximum bandwidth of a network or internet connection. It is typically undertaken by research attempts as parameter of performance evaluation by fixing the download or upload of maximum amount of data in a certain period of time. It is a measure of the frequencies, measured in hertz. In our experiment our focus remains fix at 802.11b and 802.11g with the maximum bandwidth of 11mbps and 54mbps[25].
f) Dropped Packets:
Packet loss occurs when one or more packets of data transmission is going on from source to destination across a network, but anyhow fail to reach their destination [26]. Packet loss can be a result of a number of reasons including signal deprivation over the network medium due to multi-path fading, faulty networking hardware, corrupted packets rejected intransit, faulty network drivers, channel congestion causing packet drop because of channel congestion causing packet drop or normal routing routines calls such as DSR in ad-hoc networks. Here our focus is upon calculation of number of dropped with respect to different queue size for the protocols of UDP and TCP. We know that congestion inspires packet loss. When the offered load exceeds the network capacity between few links, packets are buffered in queues[27]. Problem lies as buffers are also having limited capability, excessive congestion would lead to queue overflows, which at the end lead to packet drops. Packet loss can be actively measured by sending a set of packets from a source to a destination and comparing the number of received packets against the number of packets sent. Even few routers facilitate calculation of dropped packets as they maintain counters for calculating dropped packets and error checksum in packets[28]. Here in our experiment, we have calculated number of dropped packets at different queue length and then have analysed the results. g) Routing Protocols: This determines how the routers communicate with other routers and if not routers in between then how does end host communicate with each other. Collection of various protocols at wireless network is given as follows out of which our focus remains at AODV only: Ad-hoc: DSDV, DSR, AODV, TORA Agents are responsible for packet generation and reception. Agents work upon application layer. Agents used with our experiment are CBR(Constant Bit Rate), TCP, Sink, FTP, NULL, etc. Out of all the options available we have concentrated upon only one routing protocol that is AODV: Ad-Hoc Distance Vector Routing Protocol [29]. h) Response Time: Response time is a measure of delay between two host which is nothing but time required for a packet or datagram to travel from the source to destination and back [24]. It can also be given by time taken by each request or transaction. i.e Time taken by one transaction to complete within 1 second of time. With respect to throughput, it is just the reciprocal of throughput. In our experimentation, we have calculated the overall response time of the network for the standards 802.11b and 802.11g for different queue length.
4. OBJECTIVES: We have number of network parameters considered for the experimental analysis. Beneath we have listed the objectives for the experiment in Table I. TABLE I: Objectives of Experiment Sr. No. Objectives of Experiment 1 Implementation of test bed for TCP on 802.11b 2 Implementation of test bed for TCP on 802.11g
Karpagam Journal of Computer Science, Volume 07, Issue 03, March April 2013, pg. no. 168 to 183, ISSN 0976-2926 3 4 5
6
7
8
9
10
Implementation of test bed for UDP on 802.11b Implementation of test bed for UDP on 802.11g TCP throughput analysis with different Interface Queue Length for individual nodes and average overall network throughput analysis for 802.11b and 802.11g. UDP throughput analysis with different Interface Queue Length for individual nodes and average overall network throughput analysis for 802.11b and 802.11g. TCP Packet Drop Rate for different interface queue length for individual nodes and average overall drop rate of the network for 802.11b standard. TCP Packet Drop Rate for different interface queue length for individual nodes and average overall drop rate of the network for 802.11g standard. UDP Packet Drop Rate for different interface queue length for individual nodes and average overall drop rate of the network for 802.11b standard. UDP Packet Drop Rate for different interface queue length for individual nodes and average overall drop rate of the network for 802.11g standard.
To achieve above listed objective, we have build the simulation model with few fixed parameters with constant values where as few are varying for the entire experiment as per the experimental demand. Following given Table II is the list of configuration settings of our modelling with respect to topology specified by our experiment [30]. TABLE II: Parameter Configuration Run Parameter Value Configuration Network 500 by 500 Area No. Of 7 Nodes Queue Varying from 50 to 250 Length Queue Type Drop-tail and RED Network Uni-cast
