Computer Networks and ISDN Systems
Ethersim: a simulator for application-level performance modeling of wireless and mobile ATM networks
Introduction
The increasing deployment of wireless access technology along with the emergence of high speed integrated service networks promises to provide mobile users with ubiquitous access to multimedia information in the near future. However, transforming this vision into reality poses several problems, of which one of the most challenging is how to redesign the existing network protocols currently used in wired networks so as to allow seamless end-to-end communication over networks with both wireless and wired links. This design process requires an understanding of how network and multimedia application performance is affected by the choice of algorithms and protocols at various layers of the network hierarchy, the characteristics of the wireless links and their differences from wired links, the presence of mobile hosts, and the mobility patterns of these hosts.
Traditionally, wireless links characteristics and higher layer protocols for multimedia networks have been studied in isolation. For example, the performance of wireless links has been studied with the primary focus being on understanding and improving the performance of the physical and medium-access-control (MAC) layer protocols. Similarly, the performance of higher layer protocol suites such as ATM and TCP/IP have been studied in context of the high speed integrated service wired networks. Only recently has there been interest in studying the impact of wireless link characteristics and host mobility patterns on higher layer protocols, such as studies by various researchers on the interaction between wireless communication and the TCP protocol 11, 5, 26, 6. These studies have demonstrated that temporary disruptions in the physical connectivity due to noise and fading in wireless links, and handoffs of a mobile user from one basestation cell to another in a cellular network, can lead to long lived disruption in communication at the user level because of specific assumptions made by higher layer protocols, such as TCP. Unfortunately, since most of these studies have been conducted using actual implementations, they have not been able to extensively modify protocol policies, host mobility patterns, or link characteristics to generalize their conclusions. These complex systems are also analytically intractable, leaving simulation as often the only reasonable alternative methodology to extend analysis and obtain generalized conclusions.
In this paper we describe Ethersim, a tool designed to model application-level performance of integrated service networks with mobile hosts and wireless links. The tool grew out of a need to model and study the performance of various alternative protocols and algorithms for a wireless and mobile ATM based multimedia network called SWAN 2, 3that our research group previously developed at AT&T Bell Laboratories. In particular, we wanted to study the interaction of adaptive applications, transport layer protocols, connection rerouting schemes, and radio characteristics in a SWAN-like wireless and mobile ATM system.
Ethersim has been built using a discrete event based simulator core and incorporates models of user applications and transport, network and MAC layer protocols. It provides the capability to specify network topology and host mobility patterns. It provides the capability to specify network topology and host mobility patterns. There are five special entities in Ethersim relevant to modeling mobility and wireless communication: an air module, a map, a mover, mobile hosts, and basestations. The air module models the physical air-interface effects (e.g., RF power decay, frequency collisions, etc.). The mover is a central entity that moves the mobile hosts on the map. Ethersim allows for both random and goal-directed movements of mobile hosts, and also allows synchronized goal-directed movements to model conference room type mobility patterns. Ethersim is structured in a modular fashion to permit functionality at different levels of the protocol stack to be modified independently, thereby allowing network protocol designers to study the interaction between policies embedded in the protocols at different layers. Ethersim also includes various performance measurement and graphical user interface routines to interpret the simulation results.
By and large the available simulators that are relevant to our work fall into two broad categories: wireless link simulators and network simulators. The former focus on detailed modeling of the wireless link characteristics such as RF propagation, noise, and fading, but provide no support for modeling higher network layers. The latter focus on networking algorithms and protocols, using fixed network topologies with static hosts, and simple link models that may be adequate for wired links but are inadequate for wireless links.
Clearly, neither of these two classes of simulators is adequate for modeling wireless and mobile networks. Realizing this, various industry and academic groups developing such networks have resorted to one of three approaches to their modeling and simulation. The first approach is the use of custom developed C function libraries or C++ class libraries to create stand-alone ad hoc programs that model and simulate specific networks, such as for GSM [25]. Such custom libraries, however, cannot serve the purpose of a general simulation tool.
The second approach is to combine wireless link simulation and network simulation using software environments that either provide a mixed-domain simulator or allow multiple simulators to be interconnected using a simulator backplane. As an example, the Ptolemy [10]software environment from Berkeley provides multiple simulation domains, each corresponding to a different model of computation. Available domains include synchronous dataflow, dynamic dataflow discrete event, process networks, VHDL, etc. The Ptolemy simulator kernel orchestrates the schedulers corresponding to each of the different domains. Different parts of a system can be modeled using the domain that is well suited to it. For example, the wireless link may be modeled using one of the dataflow domains while the higher network layers may be modeled using the discrete event domain. Berkeley's Infopad wireless and mobile system project [7]used Ptolemy for some of its wireless link simulation and modeling requirements. Overall, with their roots in block diagram simulators in DSP and communications, Ptolemy and other such simulators have weaker support for network models. Also, while flexibility is a strong suite of the approach of combining multiple simulators or multiple simulation domains, it does come at a cost of efficiency.
The third approach is that of a single simulator consisting of a discrete event kernel core with entities to model various elements of a mobile and wireless network, such as the mobile hosts, the wireless link, etc. Ethersim falls into this category.
Another simulator in this category from the literature is the recent mobile wireless network simulation environment Maisie [23]used for the WAMIS mobile wireless multimedia system project at UCLA [15]. Maisie is actually a general purpose discrete event simulator capable of parallel execution on multiprocessors. It uses a specialized message-passing parallel simulation language, also called Maisie. Since Maisie has been used by various research groups for modeling wireless and mobile systems under DARPA's GloMo research program [18], a variety of models for mobile nodes, wireless channels and radios, operating system, application-specific traffic source, and network algorithm have been created by various researchers. However, reflecting their diverse origins and the inherent general-purpose nature of Maisie, these available models are not designed to all work together as a consistent library.
On a more ad hoc basis, various researchers have also used patches and tricks to coax popular network simulators, such as ns from UC Berkeley and Lawrence Berkeley National Labs (see http://www-nrg.ee.lbl.gov/ns/ and http://www-mash.cs.berkeley.edu/ns/), to model wireless channels and mobile hosts. For example, a simple wireless channel can be modeled in ns as a noisy shared channel and a CSMA based medium access protocol via an extension written by one of the co-authors (Giao Nguyen). Host mobility can be modeled in ns by using its frontend scripting language Tcl to change the network topology during the course of the simulation. In principle, more elaborate Ethersim-like notions of host location, maps and mobility traces may be programmed into ns via Tcl. However, ns and similar simulators have no direct support for mobility or shared wireless radio channels.
Finally, commercial simulators, such as OPNET, have also begun to provide modules for wireless and mobility added onto a conventional network simulator.
Ethersim, which is built on a generic discrete event kernel, is distinguished by a much general and richer variety of network components such as hosts, links, switches, and ATM and TCP/IP protocol modules that allow the modeling of a variety of mixed wired and wireless network scenarios. Instead of leaving the modeling of wireless link and the mobility entirely to the various components of the network defined by the user, Ethersim supports wireless and mobility aspects in an integral fashion as first class citizens. The notion of an air module, a map module, and a mover module are built into Ethersim, and provide efficient modeling of wireless and mobility. For example, the map and the mover together provide modeling of the geographical topology of the network and a variety of mobility patterns without burdening the user defined host models with them. Instead, an interface in terms of pre-defined mobility events is defined to the hosts.
The rest of this paper is structured as follows. In Section 2we describe the mobile and wireless network model that we assumed in designing Ethersim. In Section 3we describe the software architecture of Ethersim, and the implementation and capabilities of a few key modules. Section 4presents case studies meant to illustrate the capabilities of Ethersim. In Section 5we draw conclusions from our experience.
Section snippets
Mobile and wireless network
Ethersim uses the reference network model, shown in Fig. 1, comprised of a wired part and a wireless part. The wired part is composed of switches and wired static hosts, with point-to-point wired links connecting the hosts to switch ports, or one switch port to another switch port. This is similar to the structure of modern high-speed switched networks, such as ATM. It must be noted that the wired part of the Ethersim network model is intrinsically based on switches and point-to-point links,
Ethersim software architecture
The Ethersim software provides a modular simulator which supports all aspects of the wireless and mobile network model described in the previous section, while allowing functionality at different levels of the protocol stack to be modified. At the core of the simulator is an efficient and generic discrete event kernel. The kernel manages the scheduling of events using an event queue. The core functionality is similar to any discrete event kernel – an entity can request an event to be sent to
Case studies
In this section we will demonstrate the utility of Ethersim via two case studies. These case studies are designed to examine how wireless access and host mobility affect the achievable user and network performance, as measured by the throughput, per-packet delays and link utilization levels. In particular, we focus on the interplay between the congestion control and error recovery policies in the transport layer, the connection rerouting policies in the network layer, and the wireless radio
Conclusions
We have described Ethersim, a network simulation tool that integrates mobility and wireless links as first class citizens. Ethersim allows better modeling of mobile wireless networks than is allowed by conventional network simulators with only ad hoc support for mobility and wireless links. Ethersim addressed these problems by incorporating five special entities: an air module, a map, a mover, mobile hosts, and basestations. The air module models the physical air-interface effects such as radio
Mani Srivastava received his B.Tech. in EE from IIT Kanpur in India, and M.S. and Ph.D. from Berkeley. From 1992 through 1996 he was a Member of Technical Staff at Bell Laboratories, Murray Hill, in Networked Computing Research. Currently, he is an Assistant Professor of Electrical Engineering at UCLA. His research interests are in mobile and wireless networked computing system; low power systems; and, hardware and software synthesis for DSP and embedded systems.
References (27)
- et al.
An architecture and methodology for mobile-executed handoff in cellular ATM networks
IEEE J. Selected Area Commun.
(1994) - P. Agrawal, A. Asthana, M. Cravatts, E. Hyden, P. Krzyzanowski, P. Mishra, B. Narendran, M. Srivastava, J. Trotter, A...
- P. Agrawal, A. Asthana, M. Cravatts, E. Hyden, P. Krzyzanowski, P. Mishra, B. Narendran, M. Srivastava, J. Trotter,...
- B.A. Akyol, D.C. Cox, Rerouting for handoff in a wireless ATM network, in: Proc. ICUPC '96, September...
- A. Bakre, B.R. Badrinath, Handoff and system support for indirect TCP/IP, in: Proc. 2nd Usenix Symp. on Mobile and...
- H. Balakrishnan, S. Seshan, E. Amir, R.H. Katz, Improving TCP/IP performance over wireless networks, in: Proc. 1st...
- B. Barringer, T. Burd et al., Infopad: a system design for portable multimedia access, in: Wireless 1994, Calgary,...
- V. Bharghavan, A. Demers, S. Shenker, L. Zhang, MACAW: a media access protocol for wireless LANs, in: Proc. SIGCOMM...
- S. Biswas, A. Hopper, A connection management scheme for a mobile radio LAN, in: Proc. IEEE Int. Conf. on Personal...
- et al.
Ptolemy: a framework for simulating and prototyping heterogeneous systems
Int. J. Comput. Simulation
(1994)
Cited by (4)
Network modelling and simulation tools
2009, Simulation Modelling Practice and TheoryAn adaptive framework for QoS routing through multiple paths in ad hoc wireless networks
2003, Journal of Parallel and Distributed ComputingCitation Excerpt :We note that existing simulators for wireless mobile networks are by and large either wireless link simulators to model wireless link characteristics, or network simulators to study algorithms and protocols in a static setting. Some efforts have recently been made to combine these two classes of simulators [16,17]. They are not inadequate to model and study multihop, ad hoc wireless networks.
Tools and Techniques for Teaching and Research in Network Design and Simulation
2023, SN Computer ScienceSimulations of high performance radio LAN type 1 with ATM traffics
2000, Wireless Personal Communications
Mani Srivastava received his B.Tech. in EE from IIT Kanpur in India, and M.S. and Ph.D. from Berkeley. From 1992 through 1996 he was a Member of Technical Staff at Bell Laboratories, Murray Hill, in Networked Computing Research. Currently, he is an Assistant Professor of Electrical Engineering at UCLA. His research interests are in mobile and wireless networked computing system; low power systems; and, hardware and software synthesis for DSP and embedded systems.
Partho Mishra received the B.Tech. degree from the Indian Institute of Technology, Kharagpur, in 1988, and the M.S. and Ph.D. degrees from the University of Maryland, in 1991 and 1993, all in Computer Science. He is currently a Senior Member Technical Staff in the Networking and Distributed System Research Center at AT&T Labs-Research. His research interests include traffic management, mobile networking and packet video services. His email address is [email protected].
Prathima Agrawal is head of the networked computing technology department at AT&T Labs in Whippany, N.J., USA. She previously worked in Bell Labs as head of the networked computing research department of Murray Hill, N.J. Dr. Agrawal received her B.E. and M.E. degrees in electrical communication engineering from the Indian Institute of Science, Bangalore, India, and a Ph.D. degree in electrical engineering from the University of Southern California. Her research interests are computer networks, mobile computing, parallel processing and VLSI CAD. Dr. Agrawal is a Fellow of the IEEE.
Gia Nguyen is a Ph.D. student in Electrical Engineering and Computer Science at the University of California at Berkeley. His research interests include wireless networking and mobile computing. He received a Bachelor of Science in Computer Science and Engineering from the University of California at Los Angeles in 1994, and a Master of Science in Computer Science at the University of California at Berkeley in 1996.