ASSERT: A Wireless Networking Testbed.
- SourceAvailable from: psu.edu[show abstract] [hide abstract]
ABSTRACT: All analytical and simulation research on ad hoc wireless networks must necessarily model radio propagation using simplifying assumptions. Although it is tempting to assume that all radios have circular range, have perfect coverage in that range, and travel on a two-dimensional plane, most researchers are increasingly aware of the need to represent more realistic features, including hills, obstacles, link asymmetries, and unpredictable fading. Although many have noted the complexity of real radio propagation, and some have quantified the effect of overly simple assumptions on the simulation of ad hoc network protocols, we provide a comprehensive review of six assumptions that are still part of many ad hoc network simulation studies. In particular, we use an extensive set of measurements from a large outdoor routing experiment to demonstrate the weakness of these assumptions, and show how these assumptions cause simulation results to differ significantly from experimental results. We close with a series of recommendations for researchers, whether they develop protocols, analytic models, or simulators for ad hoc wireless networks.Simulation 07/2004; · 0.69 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: In telecommunication networks, as in many other areas of science and engineering, the proliferation of computers as research tools has resulted in the adoption of computer simulation as the most commonly used paradigm of scientific investigations. This, together with a plethora of existing simulation languages and packages, has created a popular opinion that simulation is mainly an exercise in computer programming. In new computing environments, programming can be minimized, or even fully replaced, by the manipulation of icons (representing prebuilt programming objects containing basic functional blocks of simulated systems) on a computer monitor. One can say that we have witnessed another success of modern science and technology: the emergence of wonderful and powerful tools for exploring and predicting the behavior of such complex stochastic dynamic systems as telecommunication networks. But this enthusiasm is not shared by all researchers in this area. An opinion is spreading that one cannot rely on the majority of the published results on performance evaluation studies of telecommunication networks based on stochastic simulation, since they lack credibility. Indeed, the spread of this phenomenon is so wide that one can speak about a deep crisis of credibility. In this article this claim is supported by the results of a survey of over 2200 publications on telecommunication networks in proceedings of IEEE INFOCOM and such journals as IEEE Transactions on Communications, IEEE/ACM Transactions on Networking, and Performance Evaluation Journal. The discussion focuses on two important necessary conditions of a credible simulation study: use of appropriate pseudo-random generators of independent uniformly distributed numbers, and appropriate analysis of simulation output data. Having considered their perils and pitfalls, we formulate guidelines that, if observed, could help to ensure a basic level of credibility of simulation studies of telecommunication networksIEEE Communications Magazine 02/2002; · 3.66 Impact Factor
Conference Proceeding: Characteristics, Results and Findings of IEEE 802.11 in an RF Isolated Testbed[show abstract] [hide abstract]
ABSTRACT: Simulation has been one of the most important methods for evaluating the characteristics of network protocols and applications before deployment in wireless communication networks. This paper presents the findings of a controlled and automated experimental IEEE 802.11 testbed and compares them to results of popular network simulators. The comparison shows the differences and similarities of three network simulators with the testbed using identical test scenarios. The comparison has shown that some simulators can produce misleading results under certain conditions. The discrepancy can be caused by simulators using simple and invalid models. Without realistic modelling in simulations, the evaluation of performance of mobile networks may not correlate well with performance in reality.Personal, Indoor and Mobile Radio Communications, 2007. PIMRC 2007. IEEE 18th International Symposium on; 10/2007
ASSERT: A Wireless Networking Testbed
Ehsan Nourbakhsh, Jeff Dix, Paul Johnson, Ryan Burchfield, S. Venkatesan,
Neeraj Mittal, and Ravi Prakash?
Distributed Systems Lab, Computer Science Department,
The University of Texas at Dallas 75080 USA
Abstract. As wireless networks become a critical part of home, busi-
ness and industrial infrastructure, researchers will meet these demands
by providing new networking technologies. However, these technologies
must be tested before they can be released for mainstream use. We iden-
tify the key design considerations for a wireless networking testbed as
a) accuracy b) controllability c) mobility d) repeatability e) cost effec-
tiveness f) data collection g) resource sharing h) multi-nodal capability
i) scalability. In this paper we portray how we have used coaxial cables
and our custom hardware of RF switches and programmable attenuators
to create Advanced wireleSS Enviroment Research Testbed (ASSERT),
addressing the above requirements. The created network is immune to
interference from other wireless networks, and can emulate mobility and
link deterioration. ASSERT supports various types of wireless devices,
providing researchers in academia as well as industry with the necessary
experimentation tools to validate their designed protocols and devices.
Key words: assert, Wireless, Sensor, Testbeds, Repeatability
As wireless networking is becoming more pervasive, there has been a greater
desire to develop communication hardware and protocol stacks that have a num-
ber of desirable properties like increased throughput, reduced latency, reduced
energy consumption, quality of service, security, etc. Consequently, several aca-
demic and industrial research groups are actively working towards improving
the performance of wireless networks. Due to their inherent complexity, accu-
rate theoretical analysis of the performance of large wireless networks is quite
challenging. Hence, several researchers have resorted to simulation experiments
to evaluate the performance of large wireless networks. Most simulators make a
set of simplifying assumptions about the communication medium and the com-
munication protocols [1, 2, 3, 4]. This enables them to run experiments within
a reasonable amount of time. However, sometimes these assumptions can bias
?assert was designed in collaboration with Crane Wireless Monitoring Solutions
and funded by the US Department of Defense, Defense Microelectronics Activity
2 assert: Advanced Wireless Environment Research Testbed
experiments in a significant way. It is no surprise that often the results of sim-
ulation experiments differ significantly from the actual performance of wireless
Over the last few years several research groups have initiated the develop-
ment and deployment of wireless networking testbeds such as Netbed , Kansei
, Trio , ExScal  and other testbeds at U.C. Berkeley . The underlying
assumption of all these endeavors is that experiments conducted on a testbed
composed of actual wireless devices communicating over the air will yield results
representative of performance in field deployments. Several of these testbeds are
deployed in general-purpose laboratories in academic buildings. As these labora-
tories are not shielded from external wireless interference, the experimenters have
little or no control over the environment in which experiments are conducted. As
wireless interference fluctuates in an unpredictable fashion it is also not possible
to accurately compensate for the interference. As a result, it is almost impossi-
ble for experimenters to independently reproduce the results obtained by other
Building a Faraday cage large enough to house wireless networks of non-
trivial diameter is prohibitively expensive. The other alternative of deploying
the network outdoors, sufficiently far from any interferer, is also not very at-
tractive. Outdoor deployments, unless sufficiently ruggedized, can deteriorate
quickly due to variations in temperature and humidity. They are also prone to
vandalism. Moreover, it may not be possible to conduct outdoor experiments
during inclement weather. Also, innovative ideas need to be employed to con-
duct mobile networking experiments if one does not have a lot of manpower
Based on the above discussion, one can count some general requirements for
an ideal testbed, similar to considerations that De et. al in  proposed for
a multihop testbed. The ideal testbed shall: a) accurately reflect wireless net-
work behavior (accuracy) b) give the experimenter enough control to configure
topology and environment conditions, thus eliminating uncontrolled background
noise sources (controllability) c) be able to emulate mobility of the nodes
(mobility) d) conduct experiments that are reproducible and easily repeatable
(repeatability) e) be cost effective in terms of hardware, manpower, space and
time requirements to set up, run experiments on and maintain (cost effective-
ness) f) provide necessary tools to the experimenter to collect and analyze data
(data collection) g) be able to share the available resources to conduct multi-
ple experiments without interfering with each other (resource sharing) h) have
multi-nodal capability (i.e., it will support many types of nodes) i) have the
ability to scale to a large number of nodes (scalability).
In the next sections, we will present some clear examples of challenges in
designing and building our large-scale testbed called assert(Advanced wireleSS
Environment Research Testbed). In Section 2 we show how other testbeds have
worked towards some of the above requirements. Section 3 will present some of
the main characteristics of our work and show how we are able to satisfy the
above requirements. Following this overview, we will describe the architectures
ASSERT: A Wireless Networking Testbed3
selected for both our hardware and software implementations of the assertin
Section 4. Some final conclusions and future work are discussed in Section 5.
2 Related Work
Creating an environment to test and validate new protocols and hardware designs
has been a challenge for wireless researchers. The desired environment should
be precisely controllable and the possibility of repeating the same experiment is
vital. The first group of attempts to create such environment focused on using
the antenna of unit under test (UUT), resulting in over-the-air transmissions.
Efforts such as MoteLab  and ORBIT  are examples of this category.
These testbeds do not enable the researcher to control the exposure of the UUT
to background noise and interference from other nodes. The distance between
nodes is limited to the physical placement of the devices, and mobility is not
provided by design. Although the size of the network they are able to create
is large, they are not able to partition them effectively, thus failing to address
resource sharing requirement. Noise injection and MAC filtering can be used
to create topologies, but as  pointed out repeatability and reproducibility of
results from noise injection is reduced if nodes with marginal SNR are involved.
MAC filtering will also fail to address mobility because in this method either
a packet is able to go through or is completely dropped, unlike the way actual
movement alters a signal.
Mint-m  and Mobile Emulab  address the mobility and controllabil-
ity requirements by using small robots to move the UUTs around the test area,
placing the nodes as the experimenter requests. Added attenuation between the
UUT and antenna can change the virtual distance between nodes. Pharos 
uses a similar approach of using robots, but the environment is set outdoors.
These approaches still do not eliminate the problem of exposure to background
noise or interference from other testbed devices. This would fail to address con-
trollability requirement. Also, mobility is either limited to the speed of the
robots or is not supported as a design feature.
To overcome lack of control over the environment in the aforementioned
testbeds, a second category of efforts digitize the outgoing signal of the UUTs
and use existing RF propagation models to emulate effects such as distance and
multi-path on the signal. The resulting altered signal is fed to the destination
devices. This approach provides the required controllability requirement, but
its accuracy is limited to the precision of the applied RF propagation model.
Furthermore, another limiting factor would be the available processing power.
Signal alternation requires sophisticated calculations, hence higher number of
nodes will make the processing power requirements harder to achieve, failing
to address the cost effectiveness requirement. Work done in CMU  is one
example of such testbeds.
To avoid the processing requirements and the reliance on theoretical propa-
gation models, the third category of testbeds focus on simply using coaxial cables
to connect different nodes. Attenuators and RF switches can be used to form
4 assert: Advanced Wireless Environment Research Testbed
topologies and emulate distance between nodes. MeshTest  is one example
of such set up with a design close to our work. However, their method requires
complex design for higher number of nodes. High cost of the switch matrices
used might be another obstacle to building larger networks, failing to address
cost effectiveness requirement. Our work focuses on creating a scalable testbed
using attenuators and RF switches. We isolate the nodes from each other and en-
sure that multiple experiments can run without interfering with each other. Our
design choices ensure we are able to scale to at least one thousand nodes without
unreasonable processing power requirements. We also add a generic support for
new UUTs, satisfying the multi-nodal capability requirement.
3 Design Overview
assert relies on creating virtual distances using controlled attenuation levels. If
two nodes are far from each other, the signal received on one node from the other
is attenuated by a certain calculated value to emulate distance. This attenuation
value can be calculated using simple formulae and validated through experimen-
tation, addressing the accuracy requirement. We control the attenuation value
between nodes dynamically and record our calculated values. This will allow us
to repeat the experiment exactly the same way multiple times.
Each UUT is connected to a site. This site consists of a microprocessor and
its peripherals, also a Field Programmable Gate Array (FPGA) as well as a set
of 16 attenuators and corresponding RF connectors. The site has an expandable
interface board that allows the processor to communicate with the UUT through
an interface such as RS-232 or USB. The site processor is running Linux and
is connected to an Ethernet network for network storage and communication.
Each connector can be connected to another site using a coaxial RF cable. The
main functionality of the FPGA is to set the attenuators on each RF connector
as instructed by the microprocessor and poll the RSS meter on a millisecond
basis. In order to control the environment and the way that the UUTs connect
to each other, we replace the antenna of each UUT with a coaxial connector.
This connector is connected to 16 other connectors via a set of attenuators and
RF switches. This 1-to-16 connection means we are able to add more sites and
enhance our testbed only by adding sites, addressing the scalability require-
ment. We synchronize the clock on sites through our clock distributors so that
we are able to set the attenuations on these 16 connectors, called ports, almost
instantly on all sites. This will allow us to control how the sites are “virtually”
moving in reference to each other with a high precision, addressing the mo-
bility requirement. A block diagram of a site is demonstrated in Figure 1. As
shown in Figure 2, a front-end computer called Control PC is connected to all
sites through Ethernet and is responsible for getting experiment definitiion from
user, sending control information and gathering the results.
ASSERT: A Wireless Networking Testbed5
Fig. 1. Block diagram of one site and the related RF, data and command lines
3.1 Reproducibility and Repeatability
As demonstrated in , reproducibility of experimentation can be difficult due
to inconsistency in environmental conditions. Many wireless networking testbeds
operate in schools and laboratory environments, where none-testbed devices can
interfere with an experiment. To the testbed, these devices are seen as noise. It
is important to run tests in real environments, where you can study the effect
of noise on your devices performance. But for a testbed to provide reproducible
experiments, we need the ability to control the noise around the testbed, so that
each run of an experiment will experience the same amount of noise allowing us
to make comparison across testbed experiments. As we have stated above, our
solution to this problem is to place all communication between devices on coaxial
cables. By properly shielding a wireless device, and connecting it to our coax
network, we have control over which devices can “see” each other in the network.
We can also prevent outside leakage from other wireless networks operating on
the same band.
Since we are using coaxial cable as the medium to transmit RF signal, we
would like to emulate effects the environment has on signal reception such as
multipath. To the network layer, the effects of RF propagation come down to
whether bits/packets are “lost”. By changing our attenuators according to a
fading pattern or distribution, we can emulate many real world affects on an RF
signal. Researchers will have the ability to select what attenuation patterns they
would like to use. Each site processor will calculate the attenuation for each
link due to the particular attenuation pattern and set the link’s attenuators
accordingly. The digital attenuators can be changed on a millisecond basis. All
the parameters for the link’s attenuation patterns, including random number
generator seeds are stored in a Control PC’s database, and can be retrieved
again to rerun the experiment. Using the same concept, we can also emulate
mobility. Here the key issue is a rapidly changing topology, changing neighbors,
and of course packet loss. We can do this as above by changing each link’s
attenuation dynamically to emulate the changing environment, or neighbors.
6 assert: Advanced Wireless Environment Research Testbed
3.2 Multi-nodal Capability and Data Collection
As evident in the previous sections, there are only some general expectations we
have from the unit under test. Apart from these general basic requirements, we
consider the UUT a black box which is vital for the multi-nodal capability
requirement. We expect the UUT to transmit in the frequency range our RF
equipment is designed to work. We also expect it to have an antenna that can be
replaced by a coaxial cable, so that we can connect the RF transmitter to one of
our testbed sites. The other optional requirement is to have a RS-232 interface
so that the UUT can receive commands, such as reset or load image, from the
site. This interface can be also used for the data collection mechanisms we have
provided. All logged data written by the UUT to the RS-232 serial interface is
kept on file, with timestamps of each log message, and is returned by the testbed
software as part of the experiment results.
The timestamps can be used by the experimenter to correlate data from
different units under test. For example, since all the clocks are synchronized, the
experimenter can see if a transmission by one node at a certain time has been
received at another node or not. Furthermore, we also record the RSS meter
reading of each site as well, and return the timestamped values to the user as
part of the results. In the given example, the experimenter can ensure that the
signal did go through the RF component of the sender and it did cause RSS
changes on the receiver side. This combination of correlated data can be an
important tool for the researcher, addressing the data collection requirement.
3.3 Experiment Setup and Execution
At the beginning of the experiment, the user selects if they want to create a
new experiment or repeat an already existing one. If they choose to create a new
one, they shall provide list of sites they need and the links between these sites.
They also indicate when will each link or group of links change their attenuation
pattern or value. The user is optionally prompted to upload the image to be
written to the UUT, if the UUT supports such functionality. Once a generic
image is created, specified sites are reserved for the setup stage. This will ensure
that sites are available for the experiment and no other experiment is using them.
Next step is to synchronize the clocks of the sites, and flashing the UUTs if an
image was provided. Then an XML descriptor of the timeline of the experiment
is generated. This XML file contains the ports on each site, their attenuation
pattern and the start and end time each pattern will be active. Any port not
described in this pattern is set to maximum attenuation. The XML descriptor
for each site is sent to it. After all sites receive and parse the XML descriptors,
they will start the experiment at the same time by resetting the UUTs, enabling
logging and starting to set attenuation levels on ports. The reservation is renewed
for the duration of the experiment.
Each site will report to the Control PC when the end time of the last attenu-
ation indicated in the XML descriptor has passed. Once Control PC has got the
successful termination signals from all the sites in an experiment, the Control
ASSERT: A Wireless Networking Testbed7
PC will disable logging and release the sites. The user is notified and the log files
created during the experiment are made available to them. If any of the sites
encounter a fatal error during an experiment, they will notify the Control PC.
The Control PC will then terminate the experiment early, and notify the user
that a problem has occurred.
One other method to run an experiment is to run multiple runs of the same
experiment. The same procedure will be repeated for the number of times that
user requested, and in the end the collection of all results are passed to the user.
They will be also notified if any of the runs failed. This will be a powerful tool for
experimenters, because each time the emulated environment will be replicated
precisely as the previous run.
The previous methods give the necessary support for the user to easily and
quickly set up one or multiple runs of an experiment, or repeat an existing one.
This is to address the controllability requirement mentioned earlier. It also is
addressing the cost effectiveness requirement. We are significantly reducing
the amount of time that the experimenter has to spend to create one an exper-
iment on the testbed. As we reserve the required sites during the experiment
and also set the attenuation levels to maximum on unused links, we are able to
partition the testbed and run multiple experiments in different parts of it. This
will increase the utilization of the testbed, addressing the resource sharing
4 System Architecture
4.1 Hardware Architecture
It is best to think of assert hardware as a graph(as in Figure 2), with nodes
representing sites in the testbed, and edges representing RF links between sites.
Each site consists of a custom digital board (Figure 3) and a custom RF board
(Figure 4). The digital board has a processor, memory, FPGA and serial inter-
faces where the unit under test (UUT) can connect. The processes executing on
the digital board can control the operations of the UUT, monitor the experiment,
and gather results as described in Section 4.2. The RF board connects to the
digital board and provides an interface through which the antenna port of the
UUT can connect to the RF board. The UUT interface leads to a 1 × 16 power
divider/combiner. Each output port of the power divider/combiner leads to a
programmable attenuator (controllable by the digital board) which can provide
signal attenuation between 0dB and 63.5dB, in steps of 0.25dB. The attenuators
from two different RF boards can be connected via a coaxial cable forming an
RF link between two sites. Thus the signal on this link can be attenuated in the
range 0dB to 127dB: a maximum of 63.5dB attenuation provided by each of the
two programmable attenuators on the path.
Continuing with the graph theoretic description, the digital and RF boards
together correspond to a node with a maximum of sixteen incident edges. The
8assert: Advanced Wireless Environment Research Testbed
Fig. 2. ASSERT RF Grid and control plane. Lines are coaxial cables, dotted lines are
Ethernet connections used for control and data collection.
Fig. 3. One Site Board. Each Site Board is paired with one RF board
Fig. 4. One RF Board
ASSERT: A Wireless Networking Testbed9
components on the RF board can currently support all communication bands be-
tween 720 MHz and 1125 MHz. In the near future we plan to extend the support
to the 2.4 GHz ISM band. The sites, each with a degree of sixteen, are connected
to form a mesh. By selectively maximizing signal attenuation along some links
between pairs of sites, the corresponding RF links can be removed. Similarly,
the quality of links between pairs of neighboring nodes can be manipulated by
changing the level of signal attenuation as rapidly as one db every millisecond.
4.2 Software Architecture
The assert software performs a variety of tasks, including monitoring the
testbed for faults, allocating its resources in an efficient manner, providing an
easy to use interface to users of the testbed, running user experiments on the
testbed, and gathering and reporting the results of the experiments. The soft-
ware is divided into slices, with each slice implementing a specific functionality.
We now describe the software architecture in the context of the creation and
execution of a user’s experiment.
The diagnostics slice runs both periodically and on demand to check the
integrity of RF links on the testbed. This involves selecting sites sequentially,
having them send signals along each incident link, and measuring the received
signal strength for different values of attenuation along the link. This slice reg-
isters a link between two sites if both can hear the signal generated by each
other. The quality of this link can be determined as the strength of the received
signal compared with the strength of generated signal. A run of the diagnostics
slice gives a snapshot of the topology of the testbed in terms of which links are
functional and which links are broken and need to be repaired.
The user interface slice, running on the central controller provides a graphical
user interface to the users. The network topology can be selected from a library
of topologies (like mesh, star, ring, etc.) provided by the user interface. If the
topology the user wishes to emulate is not present in the library, the user can
specify it as a graph with vertices and links between vertices. The user can also
specify the characteristics of the wireless links between vertices. Once again,
link characteristics can be specified either by selecting from a library of fading
patterns, or by providing the formulae to define the link characteristics.
Once the user has specified the desired topology, the user interface slice
invokes the experiment control slice. The experiment control slice first gathers
the current state of the testbed by querying the system state slice. The system
state slice returns the state of all the sites as well as the set of reservations
currently running on the testbed. Then the experiment control slice invokes
the topology mapper slice. The topology mapper slice computes the topology
based on user input as a subgraph of the portion of the testbed that is not
running any experiment and is thus available at the moment. The result of the
topology mapper slice’s operation is conveyed to the experiment control slice.
If the topology mapper slice is successful in finding a portion of the testbed to
run the experiment, the experiment control slice invokes the reservation slice.
The reservation slice reserves the corresponding testbed sites for the experiment.
10assert: Advanced Wireless Environment Research Testbed
These reservations are implemented as leases of finite duration. If the experiment
needs to run longer than the lease duration, the lease must be renewed prior to
Then the experiment control slice invokes the attenuator control slice at all
the reserved sites. As part of this invocation, the experiment control slice in-
forms each reserved site about the properties of the links incident on it. To
implement the desired link characteristics the sites at either end of the emu-
lated wireless link work cooperatively. Consistent with the producer-consumer
model employed by operating systems, acting as a producer each site generates
a sequence of attenuation values along with the time offset from the beginning
of the experiment when these attenuation values are to be used on a link. Act-
ing as a consumer, the FPGA on the site hardware reads these values and sets
the attenuation values accordingly once it is informed that the experiment has
started. Concurrently, the experiment control slice informs the unit under test
to start executing the experiment through the UUT control slice. Thus, as the
unit under test is running the experiment and sending and receiving messages
along the emulated wireless links, the attenuator control slice is manipulating
the characteristics of these links.
For the entire duration of an experiment running on assert, the system state
slice monitors the state of the sites. The logging slice records all error and control
messages generated by all the participating sites. The UUT logging slice records
information that the UUT writes to serial port of the site as it is running.
This data can be used for debugging the UUT by the researcher. Finally, on
the completion of the experiment the experiment control slice invokes the data
retrieval slice to gather the results of the experiment from all the participating
sites. These results, along with the UUT logs, are then conveyed to the user via
the user interface. The results can also be stored in a database associated with
After the experiment completion, the experiment control slice instructs the
reservation slice to release all the sites reserved for the experiment. It also up-
dates the system state database to reflect the availability of the newly released
sites. As a security measure, and to block access to a user’s code by other users,
the sites can be re-flashed to original configuration, wiping all user data and set-
tings from them. Similar measures are also designed in the Control PC stopping
users from having access to data of others. Furthermore, all the data includ-
ing experiment details and results can be stored on one single portable hard
drive. Also physical and remote access to the facility can be restricted, effec-
tively providing a secure environment for performing experiments that require
extra security. In a nutshell, all the software slices together can be thought of as
the operating system for assert.
The Advanced wireleSS Enviroment Research Testbed (assert) has a small foot-
print, and emulates mobility and link detoriation inside a room in a repeatable
ASSERT: A Wireless Networking Testbed 11
manner. All the experiments are controlled through front-end computers, and
network topology can be modified through a sequence of keystrokes and mouse
clicks. This takes significantly less time than physically changing the topology
in existing over-the-air wireless networking testbeds like Trio, Kansei and ExS-
cal. assert is immune to interference from other devices in the laboratory and
the environment. It will be possible for experimenters to inject noise or the de-
sired interference into the system, and observe their impact on the system being
studied. Communication between nodes in the testbed does not leak into the
environment. We perform all the signal manipulation purely in the RF domain.
This allows us to scale to higher number of nodes. Furthermore, while an RF
emulator is centralized to allow processing on a single board, our solution is
decentralized. With assert it is possible to conduct experiments in licensed
bands like the cellular service band without interfering with the services offered
by the owners of these licensed bands. Currently, network equipment vendors
are forced to test their hardwares in unpopulated areas to minimize interfer-
ence. Through sophisticated custom hardware and easy-to-use control software
assert has many valuable features that allow it to reduce the cost of testing
wireless networking protocols at scale.
Future Work: assert currently consists of forty nodes, it is designed to
scale to at least one thousand nodes without any design changes. We are con-
verting our user interface (UI) from current small Java program to a web based
application, so that the experiment set up and data gathering is done by the ex-
perimenter from their browser. The other goal is to add more preset topologies,
so that the experimenter has an extended database of already existing topologies,
while they are always able to define their own topology.
1. D. Kotz, C. Newport, R. S. Gray, J. Liu, Y. Yuan, and C. Elliott, “Experimental
evaluation of wireless simulation assumptions,” in MSWiM ’04: Proceedings of the
7th ACM international symposium on Modeling, analysis and simulation of wireless
and mobile systems, 2004.
2. K. Pawlikowski, H. Jeong, and J. Lee, “On Credibility of Simulation Studies of
Telecommunication Networks,” IEEE Communications Magazine, vol. 40, no. 1,
3. D. Cavin, Y. Sasson, and A. Schiper, “On The Accuracy of MANET Simulators,”
in Proceedings of the second ACM international workshop on Principles of Mobile
Computing. ACM New York, NY, USA, 2002.
4. R. Skehill, P. Scully, and S. McGrath, “Characteristics, Results and Findings of
IEEE 802.11 in an RF Isolated Testbed,” in 18th International Symposium on
Personal, Indoor and Mobile Radio Communications (PIMRC), 2007.
5. B. White, J. Lepreau, and S. Guruprasad, “Lowering the Barrier to Wireless and
Mobile Experimentation,” in In Proceedings of HotnetsI, 2002.
6. E. Ertin, A. Arora, R. Ramnath, V. Naik, S. Bapat, V. Kulathumani, M. Srid-
haran, H. Zhang, H. Cao, and M. Nesterenko, “Kansei: a testbed for sensing at
scale,” Proceedings of the fifth international conference on Information processing
in sensor networks, 2006.
12assert: Advanced Wireless Environment Research Testbed
7. P. Dutta, J. Hui, J. Jeong, S. Kim, C. Sharp, J. Taneja, G. Tolle, K. Whitehouse,
and D. Culler, “Trio: enabling sustainable and scalable outdoor wireless sensor
network deployments,” in Proceedings of IPSN ’06, 2006.
8. A. Arora, R. Ramnath, E. Ertin, P. Sinha, S. Bapat, V. Naik, V. Kulathumani,
H. Zhang, H. Cao, M. Sridharan, S. Kumar, N. Seddon, C. Anderson, T. Her-
man, N. Trivedi, M. Nesterenko, R. Shah, S. Kulkami, M. Aramugam, L. Wang,
M. Gouda, Y. ri Choi, D. Culler, P. Dutta, C. Sharp, G. Tolle, M. Grimmer,
B. Ferriera, and K. Parker, “Exscal: elements of an extreme scale wireless sen-
sor network,” Proceedings. 11th IEEE International Conference on Embedded and
Real-Time Computing Systems and Applications, Aug. 2005.
9. “SodaHallwireless and sensor network
10. P. De, A. Raniwala, S. Sharma, and T. cker Chiueh, “Design Considerations for a
Multihop Wireless Network Testbed,” IEEE Communications Magazine, vol. 43,
no. 10, 2005.
11. G. Werner-Allen, P. Swieskowski, and M. Welsh, “MoteLab: a wireless sensor net-
work testbed,” in Proceedings of the 4th international symposium on Information
processing in sensor networks, 2005.
12. D. Raychaudhuri, I. Seskar, M. Ott, S. Ganu, K. Ramachandran, H. Kremo, R. Sir-
acusa, H. Liu, and M. Singh, “Overview of the ORBIT radio grid testbed for eval-
uation of next-generation wireless network protocols,” Wireless Communications
and Networking Conference, 2005 IEEE, vol. 3, 2005.
13. S. Kaul, M. Gruteser, and I. Seskar, “Creating wireless multi-hop topologies on
space-constrained indoor testbeds through noise injection,” in 2nd International
Conference on Testbeds and Research Infrastructures for the Development of Net-
works and Communities (TridentCom), 2006.
14. P. De, A. Raniwala, R. Krishnan, K. Tatavarthi, J. Modi, N. A. Syed, S. Sharma,
and T. cker Chiueh, “MiNT-m: an autonomous mobile wireless experimentation
platform,” in Proceedings of the 4th international conference on Mobile Systems,
Applications and Services, 2006.
15. D. Johnson, T. Stack, R. Fish, D. M. Flickinger, L. Stoller, R. Ricci, and J. Lepreau,
“Mobile Emulab: A Robotic Wireless and Sensor Network Testbed,” Proceedings
of INFOCOM ’06, April 2006.
16. D. Stovall, N. Paine, A. Petz, J. Enderle, C. Julien, and S. Vishwanath, “Pharos:
An Application-Oriented Testbed for Heterogeneous Wireless Networking Environ-
ments,” The University of Texas at Austin, Tech. Rep. TR-UTEDGE-2009-006,
17. G. Judd, “Using Physical Layer Emulation to Understand and Improve Wireless
Networks,” Ph.D. dissertation, Carnegie Mellon University, School of Computer
Science (available as Tech Report CMU-CS-06-164), 2006.
18. T. Clancy and B. Walker, “MeshTest: Laboratory-Based Wireless Testbed for
Large Topologies,” in 3rd International Conference on Testbeds and Research In-
frastructure for the Development of Networks and Communities, (TridentCom),
19. R. Burchfield, E. Nourbakhsh, J. Dix, K. Sahu, S. Venkatesan, and R. Prakash,
“RF in the Jungle: Effect of Environment Assumptions on Wireless Experiment Re-
peatability,” in IEEE International Conference on Communications (ICC), 2009.