An Experimental Study of Network Performance Impact of Increased Latency in Software Defined Radios Thomas Schmid Oussama Sekkat Mani B.

Size: px
Start display at page:

Download "An Experimental Study of Network Performance Impact of Increased Latency in Software Defined Radios Thomas Schmid Oussama Sekkat Mani B."

Transcription

1 An Experimental Study of Network Performance Impact of Increased Latency in Software Defined Radios Thomas Schmid Oussama Sekkat Mani B. Srivastava Networked and Embedded Systems Laboratory Electrical Engineering Department, University of California, Los Angeles {thomas.schmid, osekkat, ABSTRACT Software Defined Radios are becoming more and more prevalent. Especially in the radio amateur community, Software Defined Radios are a big success. The wireless industry also has considerable interest in the dynamic reconfigurability and other advantages of Software Defined Radios. Our research focuses on the latency of Software Defined Radios and its impact on throughput in modern wireless protocols. Software Defined Radio systems often employ a bus system to transfer the samples from a radio frontend to the processor which introduces a non-negligible latency. Additionally, the signal processing calculations on general-purpose processors introduce additional latencies that are not found on conventional radios. This work concentrates on one particular Software Defined Radio system called GNU Radio, an open source Software Defined Radio application, and one of its hardware components, the Universal Software Radio Peripheral (USRP), and analyzes its receive and transmit latencies. We will use these measurements to characterize the performance impact on IEEE implementation in GNU Radio. Additionally, we present two Software Defined Radio implementations of short-range radio standards, a FSK scheme used in the Chipcon CC1000 radio, and the physical layer of IEEE We use these implementations for round trip time measurements and introduce two sample applications, a physical layer bridge between the FSK scheme and IEEE , and a dual channel receiver that receives two radio channels concurrently. Categories and Subject Descriptors C.2.1 [Computer Systems Organization]: Wireless Communication General Terms Design, Experimentation Keywords Software Defined Radio, IEEE , GNU Radio Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. WiNTECH`07, September 10, 2007, Montréal, Québec, Canada. 1. INTRODUCTION Software Defined Radio (SDR) has lately gained a lot of attention in the wireless industry as well as in academia. The recent advances in hardware design and the wide availability of very powerful processors allow us to implement some of the applications for Software Defined Radios, which were not feasible before. For example, the first software radio from the SpectrumWare Project [1] ran on a PentiumPro 133-MHz processor, and was not even able to process one AMPS cellular phone channel in realtime. In 2001, Vanu Inc. utilized similar code running on a Pentium III 1-GHz processor and was able to run seven AMPS channels in real-time. This was not achieved through better algorithms, but simply by the advances of the underlying hardware itself. Earlier this year, Vanu Inc. presented the first FCC approved multi-mode software radio base station, which combines GSM, iden, and CDMA in one device. This shows that Software Defined Radios have become a reality and today represent a very flexible and viable alternative to conventional radio systems. There is still considerable academic and industrial research concerning Software Defined Radios. In [2], Joe Mitola proposed a software radio phase space, which is a good way to compare software radios and describe where research was going on. On one side, researchers are investigating new ways to build wide band radio frontends [3]. These are generally concerned with the hardware instead of software. Nevertheless, these frontends give an immense amount of flexibility through tuneability and control knobs exposed to the software. Another area of interest is high-speed A/D converters [4], which reach the GSample/s range. The hope is that one day they will be fast enough to remove the necessity of a radio frontend altogether. Yet another area of research are the Cognitive Radios, which were introduced by J. Mitola in [5] and are part of a bigger academic field, called Cognitive Dynamic Systems [6]. Cognitive radios are a model for wireless radio nodes that can dynamically change their radio components, including their modulation scheme. The goal is to adapt to a changing wireless environment and to exploit a larger spectrum of frequencies. The finding of the Federal Communication Commission of the United States fostered these ideas, that while some frequency bands are heavily used, others are underused. Additionally, frequency usage depends heavily on time and place. Thus, a cognitive

2 radio could exploit these facts and use unused spectrum to improve its performance. The SpectrumWare project showed that general-purpose computers can be used as Software Defined Radio platforms. One might wonder why one should use such a computer when one can use more specialized hardware such as DSPs and FPGAs. Here are some of the reasons why one wishes to do so: Portability. Code written for one general purpose CPU can easily be ported to other architectures and operating systems, especially if the code is well abstracted and modularized. This gives an increased flexibility and choice in hardware, i.e., one can use the often widely available hardware necessary to fulfill the task, without being locked into a specific architecture or hardware platform. Availability of Development Tools. Development tools for general-purpose computers are more widely available, especially under open source licenses. This is very important for researchers who often need full access to the whole tool-chain in order to design new systems. Application Integration. If the Software Defined Radio code runs on the same computer as the application, then it can be better integrated into it. This can enable new models on how applications configure and use communication primitives available to them (See J. Mitola III s dissertation [5]). Many modern MAC protocols rely on specific receive and transmit latencies for different reasons, and impose deadlines by which certain actions must be done. For example in IEEE , a radio needs to wait for one Distributed Interframe Space period (DIFS, 128µs) from the time it first senses the channel to be idle, until it can actually send the message. In e, different values of DIFS are used to provide differentiated channel access to traffic of different priorities. Likewise, the acknowledgement packet in response to a data packet must be sent within a Short Interframe Space period (SIFS). Thus, a tight control over the latency and timing is necessary to meet these deadlines. In this article, we will discuss two aspects of our current research in Software Defined Radios, namely throughput and latency for packet radio systems. Michael Ismert brushes the surface of this in [7]. He shows that an increase in the buffer size of their system results in a decrease in processing overhead, but an increase in the latency of the samples coming from the A/D converter to the processor. Little attention has been paid to this inherent trade off in Software Defined Radio solutions for general purpose computing systems. In our research, we will try to quantify the increase in latency for a current commercially available Software Defined Radio hardware platform and study its impact on higher layer protocols. More specifically, we will focus on short-range wireless protocols, namely a simple FSK scheme used in the Chipcon CC1000 radio [1], and the IEEE standard. The rest of this article is structured as follows. In Section 2 we describe GNU Radio and the Universal Software Radio Peripheral (USRP) used in our research. Section 3 quantifies the sending and receiving latency of the used hardware. In Section 4 we explain the short range wireless protocols we consider in our research and briefly introduce two applications that are now possible to do with our implementations, a SDR bridge that translates between two incompatible wireless standards, and a SDR basestation that receives two radio channels at the same time. In Section 5 we evaluate the impact of measured hardware latency on the MAC protocol presented earlier and give some suggestions of changes to the hardware, software, and protocol in order to fix these problems. Section 6 describes some related work and Section 7 gives some concluding remarks. 2. Platform Our research concentrates on short-range wireless radios. More specifically, we implemented the FSK physical layer found in the Chipcon CC1000 radio chip [1] and the physical layer of IEEE For the latter, we consider a radio chip using that standard, the Chipcon CC2420 [1]. Both these standards are widely used in the sensor network community. The popular Mica2 (CC1000 radio chip), MicaZ (CC2420 radio chip) from Crossbow, or the T-Mote Sky from Moteiv all use these standards. The next two subsections will give a short introduction to GNU Radio and the USRP, two tools we use in our research. 2.1 GNU Radio GNU Radio [8] is a collection of open source software. Combining it with minimal hardware allows the construction of radios, and thus turns a typical hardware problem into multiple software problems. The main goal of GNU Radio is to facilitate the combination of signal and data processing blocks into powerful modulation, demodulation, or more complex signal processing systems. GNU Radio achieves this by providing simple signal processing primitives written in C++. By using SWIG, an interface compiler that allows easy integration of C/C++ into scripting languages, GNU Radio provides a simple interface to the signal processing blocks from within Python, a dynamic object-oriented scripting language. Thus, the Python code simply connects the signal processing blocks and allows them to run at native speed without any interpretation. Several different example applications are already written as part of GNU Radio to demonstrate its diverse signal processing tools. They range from an application that decodes HDTV pictures, an AM/FM broadcast radio en/decoder, to simple AM, FM, and PSK modulation schemes. Additionally there is an example implementation of a packet radio system using GMSK modulation and demodulation to transmit packets from one host to another. However, the problem with this system is that GNU Radio doesn t provide good support for packet based processing since it is stream oriented. A DARPA project called Adaptive Distributed Radio Open-source Intelligent Network (ADROIT) [9] attempted to change this by implementing a new primitive into GNU Radio, called the m-block [10]. This will allow a simpler implementation of block based processing and allows the annotation of samples with meta data such as timestamps. By the time of

3 Figure 1: A conventional radio has the channel sensing part very close to the PHY layer processing. Thus, latencies are small and negligible. In the ideal case, a SDR system can precompute the packet and load it into the RF frontend. But if the channel sensing is done in the CPU, it introduces a non-negligible latency that creates a blind spot that increases collisions. writing this article (Spring 2007), the m-block implementation in GNU Radio was well advanced, but not far enough to be considered for our implementations. GNU Radio by itself is not very useful, as it still needs some hardware to interface to the real world. Fortunately, GNU Radio supports several different hardware platforms [11], such as sound cards, or an RF frontend to receive different bands of the RF spectrum. The most commonly used one is the Universal Software Radio Peripheral (USRP), which we will cover in more detail in the next section. 2.2 The Universal Software Radio Peripheral Matt Ettus developed the USRP [12] as a flexible low-cost platform for software defined radios. It consists of one motherboard that holds the ADCs, DACs, and a FPGA for simple, but bandwidth consuming processing. Additionally, the FPGA is used to reduce the sampling rate such that the samples can be sent to a PC over a USB 2.0 connection. There are two Analog Devices AD9862 chips, each containing two 12-bit ADCs with up to 64 MS/s sampling rate. This allows an effective receive bandwidth of up to 32 MHz. The DACs have 14-bit resolution and up to 128 MS/s. They allow us to generate signals of up to around 50 MHz in bandwidth. From these figures we can see that the bottleneck is the USB 2.0 connection, which has a theoretic data rate limited to 480 MBit/s. In practice it is worse. A benchmark program included in GNU Radio shows, that the USB 2.0 bus can sustain about 32 MByte/s of continuous data throughput, thus limiting the transfer to a maximum of 8 MS/s of complex signals (16-bit I and Q channel). 3. Delay Measurements for the Receive and Transmit Paths Current MAC layer protocols have stringent demands on delays and latencies. Conventional radio chips have no problem to meet them, because the logic is implemented near the physical layer processing units. Thus, latency in processing and bus transfers are negligible. The following paragraphs discuss two scenarios to motivate the reason on why we need to measure and characterize the latencies in the SDR system for our research. GNU Radio does most of the processing on a generalpurpose computer and the computed samples need to be transferred over a bus system to the radio frontend. In the best case, all the samples can be precomputed and pushed Figure 2: Precomputation of a packet is not possible if it depends on the packet a radio just received. Thus, the bus system latency becomes very important and has to be as small as possible in order to minimize interframe spacings. down to the radio frontend, where a trigger can release them. This is currently not implemented in GNU Radio, but represents an optimal system that achieves the smallest latency possible. Figure 1 depicts this scenario. However, the channel sensing and sensing logic is still done on the CPU, and thus we introduce at least once the latency of the bus system. This latency inflicts a blind spot, in which someone could grab the channel and thus the SDR system would cause a collision. Figure 2 shows a different scenario. We cannot always precompute a packet the radio is about to send out if it depends on the content of the packet it just received. One such example in a current MAC layer protocol is the CTS in It must contain the receiver ID of the node the RTS came from, as well as a NAV period, which depends on the NAV period received in the RTS. Thus, the CTS needs to be calculated on the fly and we introduce at least twice the bus system latency. ACK packets have a similar problem. We can only send out an ACK once we fully decoded the message and checked it for errors, thus introducing a considerable calculation delay that needs to be characterized. To study higher layer protocol performance, one needs to know receive and transmit latencies of the USRP. In other words, we are interested in the time it takes from generating a sample on the CPU until it is sent out through the USRP, and vice versa. In order to measure this latency, we used an external oscilloscope and the computer s parallel port. The parallel port has a maximum latency of about 1 µs 1, which is, as we will see later on, much smaller than the USRP latency, and can thus be ignored. Note that the USB bus limits the sampling rate to 8MS/s for 16-bit complex samples. Thus, we only measure the latency up to that point. The latency heavily depends on the buffering introduced in the chain between the ADCs and the processing blocks running on the computer. This buffering is necessary because GNU Radio is temporally decoupled from the USRP, i.e., GNU Radio doesn t care about the physical sampling rate and processes the samples as fast as possible until its buffers are either completely full or empty. It is the USB driver that rate limits the data going to the USRP and 1 The latency here is defined as the time it takes from calling outb until the parallel port pin actually toggles. We use the latest real-time scheduling additions of the Linux kernel 2.6 and we run the processes at the highest priority (-20).

4 Figure 3: Theoretic maximum USB latency vs. measured RX latency for fusb_nblock=8 and fusb_block_size=2048. A higher sampling rate fills the buffers faster and thus decreases the latency. matches the USB transfer rate to the sampling rate. Therefore, the sampling rate influences the rate at which the buffers are filled or emptied and thus changes the latency. The USRP uses the USB in isochronous transfer mode to guarantee access to the full USB bandwidth. The downside of this mode is that the data is sent in packets over the bus and thus introduces an additional latency component. We will quantify all these latencies for streamed data as well as for burst traffic in the following paragraphs. Our test system consists of a dual Pentium IV Xeon CPU clocked at 3.75GHz and 2GB of RAM. The system runs a Linux kernel at version and we used GNU Radio at subversion revision 3941 with libusb version Receive Latency The theoretic receive latency can be calculated as follows:! =!USRP Hardware +!USB +!GNU Radio, where!usrp Hardware is the latency introduced by the USRP hardware (from the output of the antenna to the FPGA),!USB is the latency introduced by transmitting the data over the USB bus (data entering the USB bus from the FPGA until it gets handed over to the USB driver on the PC), and!gnu Radio is the latency introduced by the data processing on the computer. We can calculate the latency introduced by the USB since a USB packet is only sent when we have a sufficient amount of data collected in the USRP buffer., i.e., the smallest allowed USB packet is 512 byte, and the largest one is specified by the user by two parameters, fusb_nblocks and fusb_block_size. Thus, the USB latency is f (512, fusb _ nblocks " fusb _ block _ size)! USB =, (1) sample_ size" f s where f (x, y) depends on the amount of data in the buffer and is at least x and at most y, and f s is the sampling frequency. Since we use complex 16 bit samples, the sample size is sample_ size = 2!16bit = 4byte. With the help of an external function generator we can quantify the receive latency. The function generator produces a square wave at 1Hz. This square wave is fed into the antenna port of the USRP and at the same time into the external oscilloscope. In GNU Radio we used a very simple signal-processing block to detect the edges in the square wave transition from low to high and accordingly Figure 4: Transmit latency measurements for different sampling rates. The measurements are for busty transmissions. toggle a pin on the parallel port. This pin is connected to a second channel on the external oscilloscope where we now can measure the latency between the generated square wave and the parallel port pin s generated square wave. We set the USB transfer parameters to fusb_nblocks=8 and fusb_block_size=2048. These are reasonable values and give a good tradeoff between USB protocol overhead and USB package payload for high data rates. Figure 3 illustrates the measured receive latency for different sampling rates and compares it to the theoretic minimum and maximum USB latency calculated by using Equation 1. We give the median as well as the minimum and maximum measured latency because the collected samples do not follow a normal distribution. The behavior of the measured USRP receive latency is as expected. On average, the latency matches the maximum USB latency, i.e., each USB packet takes the maximum number of allowed bytes. The graph also shows us that there is a maximum latency that can be met with high probability. However, this latency is very high and ranges from 1 ms for a sampling rate of 8 MS/s up to 30 ms for a sampling rate of 250 ks/s. This is too large for any of the current wireless protocol standards, especially as the CPU will have to do additional processing of the samples. We will investigate in Section 5 the effects of this latency on MAC layers and give hints on how we can decrease them. 3.2 Transmit Latency The transmit latency is slightly more complicated to explain since there is a 32kByte buffer between the GNU Radio code and the USB code for the USRP. GNU Radio will immediately fill this buffer, because GNU Radio is temporally decoupled from the USRP, i.e., it is not rate limited. The temporal recoupling happens in the USB driver code. Thus, if we have a continuous transmit stream of samples, each sample has to go through this 32k buffer before it is sent to the USRP. But GNU Radio can also operate in a different way where the upper layer code generates packets, which then are converted into samples through the modulation process. These samples arrive in a burst to an empty buffer. Thus, the samples are immediately transmitted to the USRP once there are enough samples to fill a USB packet. We describe only one measurement for a continuous transmission, since the continuous streaming case is less interesting for our research. If we use a sample rate of 320kS/s it will take a complex 16-bit sample approximately 25ms to go through the 32k buffer because GNU Radio will

5 fill that buffer as fast as possible and keep it full. There is another 8 kbyte buffer in the USRP itself. This one will take an additional 6.25ms, which gives a theoretic total of 31.25ms. We measured this latency with a similar setup as with the receive latency. This time, GNU Radio generates a square wave and when the waveform goes high or low, the code toggles one of the parallel pins. The USRP output and the parallel port are connected to two different channels on an external oscilloscope, where we can measure the delay between the two waves. The median latency for a signal sampled at 320kS/s is 32.9ms, the minimum measured latency 28.9ms, and the maximum latency 36.9ms. This is what we expect given our theoretic result. The more interesting case is the burst arrival of samples. In this case, the samples do not need to go through the 32k buffer nor the 8k buffer since they should be empty. Thus, the delay can be reduced dramatically. Figure 4 depicts the measured latencies. We can see that there is a minimum latency of 200µs. We assume that this is the approximate time it takes for the data to get from user space into the USB driver since it is the same for all the sampling rates. As for the receive latency, we can define a maximum transmit latency that can be met with high probability and ranges from 240µs for 8 MS/s up to 730µs for 250 ks/s. Again, this latency does not include any preprocessing of the data which would have to be included in the latency calculations of a specific protocol implementation. As with the receive latency, the transmit latency is too high to meet the specifications of current wireless protocols. 4. Physical Layer Implementation and Applications We implemented two short-range wireless physical layer standards commonly used in sensor networks to study the impact of physical layer processing in a general purpose processor on the latency and round trip times in a packet based radio. The first physical layer uses a simple FSK modulation scheme, and the second one is the physical layer of IEEE and uses a O-QPSK modulation. Both implementations are based on the GMSK packet radio example available in GNU Radio. For more details on the implementation please see [13]. 4.1 Implementation Verification We tested both implementations with the Crossbow MicaZ and Mica2 mote that features the Chipcon CC2420 [1] and the CC1000 [1] radio transceivers. On the mote we run SOS [14], an operating system for mote-class wireless devices developed at UCLA. The network stack on the MicaZ mote is the Chipcon proprietary stack implementation that is compliant with the IEEE standard. On the Mica2 we use the default SOS stack for the Mica2 mote. We created two test scenarios to test the transmitter and the receiver code. In the first scenario, we programmed a mote to regularly send out a message. This allowed us to test the correct working of the receiver according to the standard. The messages were very simple and the pure MAC layer payload consisted of 27 bytes. Thus, the total number of bytes sent over the physical channel was 45 bytes for the MicaZ, and 37 bytes for the Mica2 (not counting the Figure 5: Measured round trip times for the conventional radios and the SDR implementations. synchronization sequence in the beginning of the message). The messages were sent at an interval of 100ms. We conducted five trials each time sending approximately 1000 messages and calculated the number of messages successfully decoded with our GNU Radio implementation. For a comparison, we also equipped a second mote with a base-station code and recorded how many messages it received. On average, the GNU Radio IEEE code received 92.8% of the messages the base-station MicaZ mote received. This is an expected number because the GNU Radio code doesn t allow any errors in the spreading sequence. The results for the GNU Radio FSK code are slightly better, 94.2%, though the native Mica2 base-station still exceeds the GNU Radio implementation s performance. The second test was done to check the transmit code of the GNU Radio implementations. We sent out the same message the mote produced with the GNU Radio system. A second computer with a second USRP and daughter-board received the messages. Additionally, a base-station mote also received the messages and we counted them as well. The messages were sent out at an interval of 500ms. On average, the GNU Radio IEEE code successfully decoded 98.6% of the messages the base-station mote received. This shows that the transmitting code works and that both the transmit and receive code can communicate with other IEEE compliant transceivers. Again, the results for the Mica2 code are very similar at 98.8%. 4.2 Round Trip Time Measurements of the SDR Protocol Implementations The round trip time of a radio is an important measure for upper layer protocols. It defines the time for a basic message exchange between two communication partners and sets limits for congestion backoffs, interframe spacing, and throughput. We compared measurements for the two SDR implementations and for the commercial radios. Figure 5 illustrates the measured times. We can see that the SDR implementation of the FSK protocol is about 8 times faster than the conventional Mica2 platform. This is not astonishing because the microcontroller of the Mica2 has to do a lot of the processing itself since the radio chip is byte oriented. Thus, the microcontroller has to search for the start of the message itself and has to spend a considerable amount of processing power, which in turn results in a higher round trip time.

6 For the IEEE implementation it is the other way around. Here, the conventional chip does most of the processing and sends the whole packet, once received, to the microcontroller. Thus, the processing is much faster and results in a 3 times shorter round trip time. Additionally, it has a very deterministic round trip time, i.e., there is not a lot of variability in the measured times. On the other side, the SDR implementation has to do a considerable amount of calculation to dispread the data and retrieve the message. This introduces a considerable delay in the communication chain. The round trip time measurements show us that we definitely need to reduce the processing delay, and that we must not simply concentrate on the hardware latency. For example, from the previous sections, we can infer that the minimal average round trip time without any data processing for the IEEE SDR implementation should be in the order of 3 ms for a sampling rate of 4 MS/s. However, we measured an average round trip time of 26.5 ms, which means that the processing, i.e., the modulation, spreading, demodulation and dispreading of the packet introduced an additional 22.5 ms. Therefore, we can improve the performance dramatically through the implementation of more efficient algorithms, and using other performance enhancements on the computer side like parallelization. 5. Implication of Physical Layer Delays on the MAC Layer, and Possible Solutions In the previous section we showed the interaction between our SDR implementations and current wireless sensing system hardware. However, this hardware does not make use of any IEEE MAC Layer standard. That being the case, we will next investigate the impact of longer latencies on such a standard, namely IEEE and discuss possibilities on how to solve these occurring problems. 5.1 Impact on IEEE The default turnaround time in IEEE is defined as 12 symbol periods for a short interframe spacing (SIFS), and 40 symbol periods for a long interframe spacing (LIFS). One symbol period in the 2.4 GHz range is 16µs, which makes the receive-transmit turnaround time 192µs or 640µs respectively. The round trip time of a packet radio is defined as RTT=2!RX+2!TX and from our round trip time measurements in Section 4.2 we know that the average round trip time of our IEEE physical layer implementation is 26.5ms. This can be translated to a receive-transmit delay of 13.25ms, which translates into 70 symbol periods. This is almost twice as long as a LIFS, and thus our current implementation could never meet the requirements for IEEE We used ns-2 [15] at version 2.30 to study the impact of a longer receive-transmit turnaround time in an extended IEEE protocol. Ns-2 includes a full implementation of IEEE , which was written by J. Zheng [16]. The scenario for our simulations included three nodes. Two of the nodes started sending constant bit rate traffic to the third node at time 7s. We let the simulation run for a total of 1000s after which we calculate the total throughput and delivery rate over that period. Figure 6: Experimental throughput study of IEEE in NS-2 for different RX-TX turnaround times. Figure 6 illustrates the result of the simulation for different receive-transmit turnaround times. We can see that under a high network load, an increase in the turnaround time decreases the total throughput of the network, because we have longer idle times between the packets. On the other hand, if the offered packet rate is lower than the network s capacity, then the larger turnaround times do not matter and the throughput is the same. 5.2 How to Solve the Problem If Added Delay Decreases Throughput In the last section we showed, that we are not able to meet the short deadlines with the current version of the USRP in current wireless protocols. We also showed relaxing these deadlines works, but significantly reduces the maximal throughput. In this section we will discuss possible solutions to these problems. In general, the solutions are of two kind: either we change the hardware/software architecture in order to comply with current protocols, or we modify protocols to cope with the restrictions given by SDR systems Hardware / Software Architecture Changes There are multiple parts in hardware and software that can be improved in order to get a better performance. The following list is not exhaustive and mentions just some of the more obvious solutions: USB: Instead of using USB2 one could change the USRP to use a different bus technology, which has a higher bandwidth and shorter latencies to the CPU. One possible candidate would be PCI-Express, which has a bandwidth of up to 8 GByte/s and could use DMA to put the samples directly into memory. But most of the general purpose bus architectures found in general purpose computers are optimized towards throughput, whereas an SDR system requires excellent latency and throughput performance. Additionally, from the measurements we presented in Section 4 we can see that even if the bus latency could be reduced to 0, we still have a large latency impact that comes from the processing of the samples on a general purpose CPU. RSSI gating and m-block: An enhancement, which is already in development in GNU Radio, is the message block (m-block). This new component will allow annotating blocks of samples with metadata like timing information. With this mechanism you will be able to instruct the USRP at what time it should send out which

7 samples. Another possibility, driven by the scenario depicted earlier on Figure 1, would be to link the USRP s FPGA to the RSSI circuit and then annotate the samples with specific RSSI characteristics, instead of timing information. Thus, the USRP sends out the samples once the measured RSSI meets these characteristics, instead of doing the channel sensing in the CPU itself. This mechanism allows for a very simple implementation of channel access methods, similar to what current wireless protocols do. But it can still not be used for packets that depend on earlier received packets since they can not be precomputed. Optimizing via precomputation: We mentioned precomputation multiple times so far as a mean to decrease latency. The goal of it is to calculate the samples that need to be sent over the air ahead of time, before the message needs to be sent. The samples get sent to the radio frontend and stored in a buffer. That way, if the radio frontend receives a interrupt from the general purpose processor, or if it detects some specific feature in the channel sensing, the samples can be sent immediately without the necessary transfer over a bus system. An extension to this scheme would be to store partial packet fragments, which always stay constant in a buffer close to the radio frontend. Thus, the processor needs only to compute the parts that change, which safes processing time as well as bus transfers. Offload often used code into hardware: Some of the most computing intensive tasks in a SDR system are the filters, and general purpose CPUs are not designed to calculate these filters efficiently. Therefore, it would make sense to push some general filter algorithms like low-pass, high-pass, band-pass, frequency translating filters, etc. down into the hardware itself. Fortunately, these filters come early in the processing stage and thus are ideal candidates to be done on the FPGA on the USRP before the data gets transferred to the computer Protocol Changes Another solution to the increase in hardware latency is to design protocols, which do not have such hard deadline constraints. Following is a non-exhaustive list of possible protocols, including changes to existing protocols, which could solve the problem: TDMA: Using a TDMA protocol would solve most of the latency problems. However, it also introduces new problems, which need to be solved such as synchronization between the participating nodes. Additionally, the m-block implementation would be needed to guarantee that samples are sent out at a specific time. This is currently not possible in GNU Radio, but should be supported in future versions. Universal Header Coding: One could imagine that the headers and ACKs use a very simple modulation scheme, and are the same through all the different protocols. This would allow implementing the demodulation in the hardware and thus a quick ACK or CTS response could be guaranteed. A similar scheme is already used today in , where the headers and beacons are sent with 1MBps, and only after the header the modulation switches to the specific speed. Delayed ACK: The increased latency is a big problem in the ACK reply and the RTS/CTS exchange. For the latter one we currently do not have a solution because we need to decode the RTS on the CPU in order to get the address and timing information. Additionally, the RTS reserves the channel for a certain amount of time and no one can use it for that time, except the node addressed in the RTS. Thus, we have to accept the latency increase. For the ACKs, we have some more liberty. For example, we can delay them to a later point in time. A hypothetical protocol could look like this: assume that we have a similar protocol as IEEE , i.e., the channel needs to be clear for at least a DIFS period before one can try to acquire it. Unfortunately, the SDR implementation cannot reply to a message until 1.5! DIFS, i.e., if someone else tries to acquire the channel in order to send a message, the ACK cannot be sent immediately after the message. Therefore, if no one needs the channel (low channel utilization), we can afford the latency and send the ACK after 1.5! DIFS. Else, if someone acquired the channel to send a message, the ACK-ing node waits until the end of the transmission and then grabs the channel at 0.5! DIFS, i.e., he has priority over all the other nodes. Note that we did not do any formal analysis of such a protocol. It is simply an idea of how one might solve the latency problem and certainly needs much more work to determine its feasibility. 6. Related Work S. Valentin describes in [17] the delays measured in their GNU Radio testbed. He states that the USB bus transfer and USRP calculations can be neglected. As we showed in Section 3 this is not the case. Unfortunately, [17] fails to mention the sampling rate at which they run the USRP. However, if we assume that they run it at 8MS/s, then we can find from our measurements the total USB latency as 2"(RX Latency + TX Latency) = 1.6ms. This is close to the 1.4 ms delay Valentin found four their setup. The SpectrumWare project had its own interface card, the GuPPI [7]. This card used the PCI bus and DMA for a fast memory access. The minimum input latency they could achieve was a little bit over 1 ms. Additionally, the GuPPI project used some virtual memory tricks in order to decrease the OS overhead for memory access. It remains to be investigated if such tricks could also decrease the latency in GNU Radio. The CalRADIO 1 [18] from UCSD takes a different approach. Their goal is to provide a general development platform for the layers 2-7 of b. Their physical layer is fixed and implemented in the Intersil (Conexant) Prism chip. Everything else is done in software on a C5471 Texas Instruments DSP, which contains an ARM processor. There are many other SDR platforms available commercially, such as the XMC-3321 from Spectrum Signal, or in academia, like the MIFE system from EPFL[19]. Unfortunately, the discussion of each of these systems available is out of scope for this publication. 7. Conclusion SDR platforms have emerged as a flexible alternative to conventional radios, but their very nature of pushing the physical layer processing into general purpose processors

8 increases the latency of available interconnects. This latency needs to be characterized and should be as small as possible. We conducted such latency measurements for the receive and transmit path of the USRP in order to study the impacts on MAC layer protocols. We found that the minimal receive latency is about 600µs and the minimal transmit latency about 200µs. We found that these latencies are already too long for the IFS requirements of modern MAC protocols. However, they do not include any calculation in the processor itself, which will increase that latency even more. In order to study the impact of larger latencies, we implemented two short-range radio standards, the physical layer of IEEE and a simple FSK scheme. These two implementations allowed us to make an assessment on the latency including modulation/demodulation and packet processing. We found in our experimental analysis that the round trip time of our IEEE SDR implementation is 25ms on average and 50ms maximum. This is slow compared to a conventional radio chip, which achieves a round trip time of 8 ms, and shows that latency is a big issue. As a consequence, we studied the impact on a MAC layer considering an increase in latency. We found that the throughput of IEEE decreases by about 30% under a heavy network load. Finally, we presented ideas on how to solve this problem. The solutions we gave are two fold, based on changes in hardware and software in order to meet latencies of existing MAC protocols, or based on modifications and adaptation of the used wireless protocols to make them cope with the problems introduced by SDR systems. The implementation of the two physical layer implementations in GNU Radio allow us to do further research on different aspects in Software Defined Radio. The implementation of two sample applications, a dualchannel radio and a physical layer bridge, give a first impression on the potential of Software Defined Radios in wireless sensing systems. The field of cognitive radios is growing rapidly, and we plan to integrate cognitive radio principles, such as dynamic channel and modulation selection, into our software solution. 8. REFERENCES [1] D. Tennenhouse, V. Bose. The SpectrumWare Approach to Wireless Signal Processing, Wireless Network Journal, 1996 [2] J. Mitola. Software Radio Architecture: a Mathematical Perspective, IEEE Journal on Selected Areas in Communications 17 (4) (1999) [3] S. Lang, B. Daneshrad. Design and Implementation of a 5.25 GHz Radio Transceiver for a MIMO Testbed, Wireless Communications and Networking Conference, [4] K. Lundberg. High-Speed Analog-to-Digital Converter Survey, unpublished, [5] J. Mitola. Cognitive Radio, Licentiate Thesis, Dept. of Teleinformatics, Royal Institute of Technology, Sweden, [6] S. Haykin. Cognitive Dynamic Systems, under preparation, [7] M. Ismert. Making Commodity PCs Fit for Signal Processing, USENIX, [8] GNU Radio. Mai [9] ADROIT website, Mai [10] ADROIT: GNU Radio Architectural Changes, Mai [11] Supported GNU Radio hardware, GnuRadioHardware, Mai [12] Ettus Research LLC, Universal Software Radio Peripheral, Mai 2007 [13] Mai 2007 [14] C. Han, R. Rengaswamy, R. Shea, E. Kohler, M. Srivastava. Sos: A dynamic Operating System for Sensor Networks, SenSys [15] The Network Simulator - ns-2, Mai [16] J. Zheng and M. Lee. A Comprehensive Performance Study of IEEE , IEEE Press Book, [17] S. Valentin and H. von Malm and H. Karl. Evaluating the GNU Software Radio Platform for Wireless Testbeds, Technical Report TR-RI , February 2006 [18] UCSD, CalRadio1, Mai [19] MIFE, Mai 2007.

An Experimental Study of Network Performance Impact of Increased Latency in SDR

An Experimental Study of Network Performance Impact of Increased Latency in SDR An Experimental Study of Network Performance Impact of Increased Latency in SDR Thomas Schmid Oussama Sekkat Mani B. Srivastava - Wintech workshop was started with the Keynote from Eric Blossom on GNU

More information

Evaluating the GNU Software Radio platform for wireless testbeds

Evaluating the GNU Software Radio platform for wireless testbeds University of Paderborn Computer Networks Group Evaluating the GNU Software Radio platform for wireless testbeds Stefan Valentin, Holger von Malm, Holger Karl {stefanv holgervm holger.karl}@upb.de February

More information

Wireless Communications

Wireless Communications 4. Medium Access Control Sublayer DIN/CTC/UEM 2018 Why do we need MAC for? Medium Access Control (MAC) Shared medium instead of point-to-point link MAC sublayer controls access to shared medium Examples:

More information

Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A Single Transceiver

Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A Single Transceiver Multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A Single Transceiver Jungmin So Dept. of Computer Science, and Coordinated Science Laboratory University of Illinois

More information

Guide to Wireless Communications, Third Edition. Objectives

Guide to Wireless Communications, Third Edition. Objectives Guide to Wireless Communications, Third Edition Chapter 7 Low-Speed Wireless Local Area Networks Objectives Describe how WLANs are used List the components and modes of a WLAN Describe how an RF WLAN works

More information

Wireless Local Area Networks (WLANs)) and Wireless Sensor Networks (WSNs) Computer Networks: Wireless Networks 1

Wireless Local Area Networks (WLANs)) and Wireless Sensor Networks (WSNs) Computer Networks: Wireless Networks 1 Wireless Local Area Networks (WLANs)) and Wireless Sensor Networks (WSNs) Computer Networks: Wireless Networks 1 Wireless Local Area Networks The proliferation of laptop computers and other mobile devices

More information

Simplifying FPGA Design with A Novel Network-on-Chip Architecture

Simplifying FPGA Design with A Novel Network-on-Chip Architecture Simplifying FPGA Design with A Novel Network-on-Chip Architecture ABSTRACT John Malsbury Ettus Research 1043 N Shoreline Blvd Suite 100 +1 (650) 967-2870 john.malsbury@ettus.com As wireless communications

More information

15-441: Computer Networking. Wireless Networking

15-441: Computer Networking. Wireless Networking 15-441: Computer Networking Wireless Networking Outline Wireless Challenges 802.11 Overview Link Layer Ad-hoc Networks 2 Assumptions made in Internet Host are (mostly) stationary Address assignment, routing

More information

Local Area Networks NETW 901

Local Area Networks NETW 901 Local Area Networks NETW 901 Lecture 4 Wireless LAN Course Instructor: Dr.-Ing. Maggie Mashaly maggie.ezzat@guc.edu.eg C3.220 1 Contents What is a Wireless LAN? Applications and Requirements Transmission

More information

Advanced Computer Networks WLAN

Advanced Computer Networks WLAN Advanced Computer Networks 263 3501 00 WLAN Patrick Stuedi Spring Semester 2014 1 Oriana Riva, Department of Computer Science ETH Zürich Last week Outlook Medium Access COPE Short Range Wireless Networks:

More information

6.9. Communicating to the Outside World: Cluster Networking

6.9. Communicating to the Outside World: Cluster Networking 6.9 Communicating to the Outside World: Cluster Networking This online section describes the networking hardware and software used to connect the nodes of cluster together. As there are whole books and

More information

1. IEEE and ZigBee working model.

1. IEEE and ZigBee working model. ZigBee SoCs provide cost-effective solutions Integrating a radio transceiver, data processing unit, memory and user-application features on one chip combines high performance with low cost Khanh Tuan Le,

More information

The WINLAB Cognitive Radio Platform

The WINLAB Cognitive Radio Platform The WINLAB Cognitive Radio Platform IAB Meeting, Fall 2007 Rutgers, The State University of New Jersey Ivan Seskar Software Defined Radio/ Cognitive Radio Terminology Software Defined Radio (SDR) is any

More information

3.1. Introduction to WLAN IEEE

3.1. Introduction to WLAN IEEE 3.1. Introduction to WLAN IEEE 802.11 WCOM, WLAN, 1 References [1] J. Schiller, Mobile Communications, 2nd Ed., Pearson, 2003. [2] Martin Sauter, "From GSM to LTE", chapter 6, Wiley, 2011. [3] wiki to

More information

Data Communications. Data Link Layer Protocols Wireless LANs

Data Communications. Data Link Layer Protocols Wireless LANs Data Communications Data Link Layer Protocols Wireless LANs Wireless Networks Several different types of communications networks are using unguided media. These networks are generally referred to as wireless

More information

Lesson 2-3: The IEEE x MAC Layer

Lesson 2-3: The IEEE x MAC Layer Module 2: Establishing Wireless Connectivity Lesson 2-3: The IEEE 802.11x MAC Layer Lesson Overview This lesson describes basic IEEE 802.11x MAC operation, beginning with an explanation of contention schemes

More information

Latency Analysis in GNU Radio/USRP-based Software Radio Platforms

Latency Analysis in GNU Radio/USRP-based Software Radio Platforms 2013 IEEE Military Communications Conference Latency Analysis in GNU Radio/USRP-based Software Radio Platforms Nguyen B.Truong DASAN NETWORKS Corp. Seoul, Gyeonggi-do, 463-400 South Korea nguyentb@postech.ac.kr

More information

Solving the Data Transfer Bottleneck in Digitizers

Solving the Data Transfer Bottleneck in Digitizers Solving the Data Transfer Bottleneck in Digitizers With most modern PC based digitizers and data acquisition systems a common problem is caused by the fact that the ADC technology usually runs in advance

More information

original standard a transmission at 5 GHz bit rate 54 Mbit/s b support for 5.5 and 11 Mbit/s e QoS

original standard a transmission at 5 GHz bit rate 54 Mbit/s b support for 5.5 and 11 Mbit/s e QoS IEEE 802.11 The standard defines a wireless physical interface and the MAC layer while LLC layer is defined in 802.2. The standardization process, started in 1990, is still going on; some versions are:

More information

Department of Electrical and Computer Systems Engineering

Department of Electrical and Computer Systems Engineering Department of Electrical and Computer Systems Engineering Technical Report MECSE-6-2006 Medium Access Control (MAC) Schemes for Quality of Service (QoS) provision of Voice over Internet Protocol (VoIP)

More information

Wireless Local Area Networks. Networks: Wireless LANs 1

Wireless Local Area Networks. Networks: Wireless LANs 1 Wireless Local Area Networks Networks: Wireless LANs 1 Wireless Local Area Networks The proliferation of laptop computers and other mobile devices (PDAs and cell phones) created an obvious application

More information

Multiple Access Links and Protocols

Multiple Access Links and Protocols Multiple Access Links and Protocols Two types of links : point-to-point PPP for dial-up access point-to-point link between Ethernet switch and host broadcast (shared wire or medium) old-fashioned Ethernet

More information

Wireless# Guide to Wireless Communications. Objectives

Wireless# Guide to Wireless Communications. Objectives Wireless# Guide to Wireless Communications Chapter 7 Low-Speed Wireless Local Area Networks Objectives Describe how WLANs are used List the components and modes of a WLAN Describe how an RF WLAN works

More information

Ettus Research Update

Ettus Research Update Ettus Research Update Matt Ettus Ettus Research GRCon13 Outline 1 Introduction 2 Recent New Products 3 Third Generation Introduction Who am I? Core GNU Radio contributor since 2001 Designed

More information

Wireless Local Area Networks (WLANs) and Wireless Sensor Networks (WSNs) Primer. Computer Networks: Wireless LANs

Wireless Local Area Networks (WLANs) and Wireless Sensor Networks (WSNs) Primer. Computer Networks: Wireless LANs Wireless Local Area Networks (WLANs) and Wireless Sensor Networks (WSNs) Primer 1 Wireless Local Area Networks (WLANs) The proliferation of laptop computers and other mobile devices (PDAs and cell phones)

More information

Wireless LANs. ITS 413 Internet Technologies and Applications

Wireless LANs. ITS 413 Internet Technologies and Applications Wireless LANs ITS 413 Internet Technologies and Applications Aim: Aim and Contents Understand how IEEE 802.11 wireless LANs work Understand what influences the performance of wireless LANs Contents: IEEE

More information

LXRS and LXRS+ Wireless Sensor Protocol

LXRS and LXRS+ Wireless Sensor Protocol LORD TECHNICAL NOTE LXRS and LXRS+ Wireless Sensor Protocol Using LXRS and LXRS+ For Long-Term Monitoring and High Bandwidth Test and Measurement Introduction LORD Sensing has developed and deployed two

More information

Enabling MAC Protocol Implementations on Software-Defined Radios

Enabling MAC Protocol Implementations on Software-Defined Radios Enabling MAC Protocol Implementations on Software-Defined Radios George Nychis, Thibaud Hottelier, Zhuocheng Yang, Srinivasan Seshan, Peter Steenkiste Carnegie Mellon University Abstract Over the past

More information

Mobile Communications Chapter 7: Wireless LANs

Mobile Communications Chapter 7: Wireless LANs Characteristics IEEE 802.11 PHY MAC Roaming IEEE 802.11a, b, g, e HIPERLAN Bluetooth Comparisons Prof. Dr.-Ing. Jochen Schiller, http://www.jochenschiller.de/ MC SS02 7.1 Comparison: infrastructure vs.

More information

04/11/2011. Wireless LANs. CSE 3213 Fall November Overview

04/11/2011. Wireless LANs. CSE 3213 Fall November Overview Wireless LANs CSE 3213 Fall 2011 4 November 2011 Overview 2 1 Infrastructure Wireless LAN 3 Applications of Wireless LANs Key application areas: LAN extension cross-building interconnect nomadic access

More information

Wireless Challenges : Computer Networking. Overview. Routing to Mobile Nodes. Lecture 25: Wireless Networking

Wireless Challenges : Computer Networking. Overview. Routing to Mobile Nodes. Lecture 25: Wireless Networking Wireless Challenges 15-441: Computer Networking Lecture 25: Wireless Networking Force us to rethink many assumptions Need to share airwaves rather than wire Don t know what hosts are involved Host may

More information

LANCOM Techpaper IEEE n Indoor Performance

LANCOM Techpaper IEEE n Indoor Performance Introduction The standard IEEE 802.11n features a number of new mechanisms which significantly increase available bandwidths. The former wireless LAN standards based on 802.11a/g enable physical gross

More information

4.3 IEEE Physical Layer IEEE IEEE b IEEE a IEEE g IEEE n IEEE 802.

4.3 IEEE Physical Layer IEEE IEEE b IEEE a IEEE g IEEE n IEEE 802. 4.3 IEEE 802.11 Physical Layer 4.3.1 IEEE 802.11 4.3.2 IEEE 802.11b 4.3.3 IEEE 802.11a 4.3.4 IEEE 802.11g 4.3.5 IEEE 802.11n 4.3.6 IEEE 802.11ac,ad Andreas Könsgen Summer Term 2012 4.3.3 IEEE 802.11a Data

More information

Packet-Level Diversity From Theory to Practice: An based Experimental Investigation

Packet-Level Diversity From Theory to Practice: An based Experimental Investigation Packet-Level Diversity From Theory to Practice: An 802.11- based Experimental Investigation E. Vergetis, E. Pierce, M. Blanco and R. Guérin University of Pennsylvania Department of Electrical & Systems

More information

Wireless Communication and Networking CMPT 371

Wireless Communication and Networking CMPT 371 Wireless Communication and Networking CMPT 371 Wireless Systems: AM, FM Radio TV Broadcast Satellite Broadcast 2-way Radios Cordless Phones Satellite Links Mobile Telephony Systems Wireless Local Loop

More information

Wireless MACs: MACAW/802.11

Wireless MACs: MACAW/802.11 Wireless MACs: MACAW/802.11 Mark Handley UCL Computer Science CS 3035/GZ01 Fundamentals: Spectrum and Capacity A particular radio transmits over some range of frequencies; its bandwidth, in the physical

More information

CHAPTER 4 CROSS LAYER INTERACTION

CHAPTER 4 CROSS LAYER INTERACTION 38 CHAPTER 4 CROSS LAYER INTERACTION The cross layer interaction techniques used in the lower layers of the protocol stack, solve the hidden and exposed terminal problems of wireless and ad hoc networks.

More information

Wireless Networked Systems

Wireless Networked Systems Wireless Networked Systems CS 795/895 - Spring 2013 Lec #6: Medium Access Control QoS and Service Differentiation, and Power Management Tamer Nadeem Dept. of Computer Science Quality of Service (802.11e)

More information

Mohammad Hossein Manshaei 1393

Mohammad Hossein Manshaei 1393 Mohammad Hossein Manshaei manshaei@gmail.com 1393 1 An Analytical Approach: Bianchi Model 2 Real Experimentations HoE on IEEE 802.11b Analytical Models Bianchi s Model Simulations ns-2 3 N links with the

More information

An Efficient Scheduling Scheme for High Speed IEEE WLANs

An Efficient Scheduling Scheme for High Speed IEEE WLANs An Efficient Scheduling Scheme for High Speed IEEE 802.11 WLANs Juki Wirawan Tantra, Chuan Heng Foh, and Bu Sung Lee Centre of Muldia and Network Technology School of Computer Engineering Nanyang Technological

More information

An IEEE MAC Software Defined Radio Implementation for Experimental Wireless Communications and Networking Research

An IEEE MAC Software Defined Radio Implementation for Experimental Wireless Communications and Networking Research An IEEE 82.11 MAC Software Defined Radio Implementation for Experimental Wireless Communications and Networking Research Juan R. Gutierrez-Agullo, Baldomero Coll-Perales and Javier Gozalvez Uwicore, Ubiquitous

More information

Medium Access Control. MAC protocols: design goals, challenges, contention-based and contention-free protocols

Medium Access Control. MAC protocols: design goals, challenges, contention-based and contention-free protocols Medium Access Control MAC protocols: design goals, challenges, contention-based and contention-free protocols 1 Why do we need MAC protocols? Wireless medium is shared Many nodes may need to access the

More information

Wireless LAN -Architecture

Wireless LAN -Architecture Wireless LAN -Architecture IEEE has defined the specifications for a wireless LAN, called IEEE 802.11, which covers the physical and data link layers. Basic Service Set (BSS) Access Point (AP) Distribution

More information

Implementation and Analysis of Large Receive Offload in a Virtualized System

Implementation and Analysis of Large Receive Offload in a Virtualized System Implementation and Analysis of Large Receive Offload in a Virtualized System Takayuki Hatori and Hitoshi Oi The University of Aizu, Aizu Wakamatsu, JAPAN {s1110173,hitoshi}@u-aizu.ac.jp Abstract System

More information

PC/104 Test-Bed for Software GPS Receiver (SGR) and Software Defined Radio (SDR) Applications

PC/104 Test-Bed for Software GPS Receiver (SGR) and Software Defined Radio (SDR) Applications PC/104 Test-Bed for Software GPS Receiver (SGR) and Software Defined Radio (SDR) Applications Frank Carpenter, NAVSYS Corporation 14960 Woodcarver Road Colorado Springs, CO 80921 (719) 481-4877 x137 (719)

More information

Chapter 12 Multiple Access 12.1

Chapter 12 Multiple Access 12.1 Chapter 12 Multiple Access 12.1 Copyright The McGraw-Hill Companies, Inc. Permission required for reproduction or display. 12.2 Figure 12.1 Data link layer divided into two functionality-oriented sublayers

More information

Title: Using low-power dual-port for inter processor communication in next generation mobile handsets

Title: Using low-power dual-port for inter processor communication in next generation mobile handsets Title: Using low-power dual-port for inter processor communication in next generation mobile handsets Abstract: The convergence of mobile phones and other consumer-driven devices such as PDAs, MP3 players,

More information

DOMINO: A System to Detect Greedy Behavior in IEEE Hotspots

DOMINO: A System to Detect Greedy Behavior in IEEE Hotspots DOMINO: A System to Detect Greedy Behavior in IEEE 802.11 Hotspots By Maxim Raya, Jean-Pierre Hubaux, Imad Aad Laboratory for computer Communications and Applications(LCA) School of Computer and Communication

More information

standard. Acknowledgement: Slides borrowed from Richard Y. Yale

standard. Acknowledgement: Slides borrowed from Richard Y. Yale 802.11 standard Acknowledgement: Slides borrowed from Richard Y. Yang @ Yale IEEE 802.11 Requirements Design for small coverage (e.g. office, home) Low/no mobility High data rate applications Ability to

More information

Department of EECS - University of California at Berkeley EECS122 - Introduction to Communication Networks - Spring 2005 Final: 5/20/2005

Department of EECS - University of California at Berkeley EECS122 - Introduction to Communication Networks - Spring 2005 Final: 5/20/2005 Name: SID: Department of EECS - University of California at Berkeley EECS122 - Introduction to Communication Networks - Spring 2005 Final: 5/20/2005 There are 10 questions in total. Please write your SID

More information

Investigating MAC-layer Schemes to Promote Doze Mode in based WLANs

Investigating MAC-layer Schemes to Promote Doze Mode in based WLANs Investigating MAC-layer Schemes to Promote Doze Mode in 802.11-based WLANs V. Baiamonte and C.-F. Chiasserini CERCOM - Dipartimento di Elettronica Politecnico di Torino Torino, Italy Email: baiamonte,chiasserini

More information

MAC. Fall Data Communications II 1

MAC. Fall Data Communications II 1 802.11 MAC Fall 2005 91.564 Data Communications II 1 RF Quality (ACK) Fall 2005 91.564 Data Communications II 2 Hidden Terminal (RTS/CTS) Fall 2005 91.564 Data Communications II 3 MAC Coordination Functions

More information

Getting Connected (Chapter 2 Part 4) Networking CS 3470, Section 1 Sarah Diesburg

Getting Connected (Chapter 2 Part 4) Networking CS 3470, Section 1 Sarah Diesburg Getting Connected (Chapter 2 Part 4) Networking CS 3470, Section 1 Sarah Diesburg Five Problems Encoding/decoding Framing Error Detection Error Correction Media Access Five Problems Encoding/decoding Framing

More information

IEEE , Token Rings. 10/11/06 CS/ECE UIUC, Fall

IEEE , Token Rings. 10/11/06 CS/ECE UIUC, Fall IEEE 802.11, Token Rings 10/11/06 CS/ECE 438 - UIUC, Fall 2006 1 Medium Access Control Wireless channel is a shared medium Need access control mechanism to avoid interference Why not CSMA/CD? 10/11/06

More information

Data and Computer Communications. Chapter 13 Wireless LANs

Data and Computer Communications. Chapter 13 Wireless LANs Data and Computer Communications Chapter 13 Wireless LANs Wireless LAN Topology Infrastructure LAN Connect to stations on wired LAN and in other cells May do automatic handoff Ad hoc LAN No hub Peer-to-peer

More information

Power-efficient Communication Protocol for Social Networking Tags for Visually Impaired

Power-efficient Communication Protocol for Social Networking Tags for Visually Impaired Power-efficient Communication Protocol for Social Networking Tags for Visually Impaired Problem Social Networking Tags System for Visually Impaired is an project aims to utilize electronic id technology

More information

Computer Networks. Wireless LANs

Computer Networks. Wireless LANs Computer Networks Wireless LANs Mobile Communication Technology according to IEEE (examples) Local wireless networks WLAN 802.11 Personal wireless nw WPAN 802.15 WiFi 802.11a 802.11b 802.11h 802.11i/e/

More information

Wireless and Mobile Networks

Wireless and Mobile Networks Wireless and Mobile Networks Raj Jain Washington University in Saint Louis Saint Louis, MO 63130 Jain@wustl.edu Audio/Video recordings of this lecture are available on-line at: http://www.cse.wustl.edu/~jain/cse473-11/

More information

MSIT 413: Wireless Technologies Week 8

MSIT 413: Wireless Technologies Week 8 MSIT 413: Wireless Technologies Week 8 Michael L. Honig Department of EECS Northwestern University November 2017 The Multiple Access Problem How can multiple mobiles access (communicate with) the same

More information

Introduction to Real-Time Communications. Real-Time and Embedded Systems (M) Lecture 15

Introduction to Real-Time Communications. Real-Time and Embedded Systems (M) Lecture 15 Introduction to Real-Time Communications Real-Time and Embedded Systems (M) Lecture 15 Lecture Outline Modelling real-time communications Traffic and network models Properties of networks Throughput, delay

More information

Module 6: INPUT - OUTPUT (I/O)

Module 6: INPUT - OUTPUT (I/O) Module 6: INPUT - OUTPUT (I/O) Introduction Computers communicate with the outside world via I/O devices Input devices supply computers with data to operate on E.g: Keyboard, Mouse, Voice recognition hardware,

More information

HPCC Random Access Benchmark Excels on Data Vortex

HPCC Random Access Benchmark Excels on Data Vortex HPCC Random Access Benchmark Excels on Data Vortex Version 1.1 * June 7 2016 Abstract The Random Access 1 benchmark, as defined by the High Performance Computing Challenge (HPCC), tests how frequently

More information

Lecture 16: QoS and "

Lecture 16: QoS and Lecture 16: QoS and 802.11" CSE 123: Computer Networks Alex C. Snoeren HW 4 due now! Lecture 16 Overview" Network-wide QoS IntServ DifServ 802.11 Wireless CSMA/CA Hidden Terminals RTS/CTS CSE 123 Lecture

More information

Optical Data Interface ODI-2 Transport Layer Preliminary Specification. Revision Date

Optical Data Interface ODI-2 Transport Layer Preliminary Specification. Revision Date Optical Data Interface O-2 Transport Layer Preliminary Specification Revision Date 171002 2 O 3-part Specification O-2.1: High-Speed Formats 8 to 16 bit data formats Packing Methods Optimized for SDR &

More information

Strengthening Unlicensed Band Wireless Backhaul

Strengthening Unlicensed Band Wireless Backhaul be in charge Strengthening Unlicensed Band Wireless Backhaul Use TDD/TDMA Based Channel Access Mechanism WHITE PAPER Strengthening Unlicensed Band Wireless Backhaul: Use TDD/TDMA Based Channel Access Mechanism

More information

2. REAL-TIME CONTROL SYSTEM AND REAL-TIME NETWORKS

2. REAL-TIME CONTROL SYSTEM AND REAL-TIME NETWORKS 2. REAL-TIME CONTROL SYSTEM AND REAL-TIME NETWORKS 2.1 Real-Time and Control Computer based digital controllers typically have the ability to monitor a number of discrete and analog inputs, perform complex

More information

Configuring the Wireless Parameters (CPE and WBS)

Configuring the Wireless Parameters (CPE and WBS) Configuring the Wireless Parameters (CPE and WBS) CHAPTERS 1. Configure Basic Wireless Parameters 2. Configure Wireless Client Parameters 3. Configure Wireless AP Parameters 4. Configure Multi-SSID 5.

More information

standards like IEEE [37], IEEE [38] or IEEE [39] do not consider

standards like IEEE [37], IEEE [38] or IEEE [39] do not consider Chapter 5 IEEE 802.15.4 5.1 Introduction Wireless Sensor Network(WSN) is resource constrained network developed specially targeting applications having unattended network for long time. Such a network

More information

Lecture 23 Overview. Last Lecture. This Lecture. Next Lecture ADSL, ATM. Wireless Technologies (1) Source: chapters 6.2, 15

Lecture 23 Overview. Last Lecture. This Lecture. Next Lecture ADSL, ATM. Wireless Technologies (1) Source: chapters 6.2, 15 Lecture 23 Overview Last Lecture ADSL, ATM This Lecture Wireless Technologies (1) Wireless LAN, CSMA/CA, Bluetooth Source: chapters 6.2, 15 Next Lecture Wireless Technologies (2) Source: chapter 16, 19.3

More information

Avoid Bottlenecks Using PCI Express-Based Embedded Systems

Avoid Bottlenecks Using PCI Express-Based Embedded Systems Avoid Bottlenecks Using PCI Express-Based Embedded Systems Implementing efficient data movement is a critical element in high-performance embedded systems, and the advent of PCI Express has presented us

More information

RF OVER ETHERNET FOR WIRELESS INFRASTRUCTURE

RF OVER ETHERNET FOR WIRELESS INFRASTRUCTURE RF OVER ETHERNET FOR WIRELESS INFRASTRUCTURE Gerald Britton, Byron Kubert, and John Chapin (Vanu, Inc., Cambridge, MA, USA) {gbritton,bkubert,jchapin}@vanu.com ABSTRACT Existing interconnects between RF

More information

Design and Deployment Considerations for High Performance MIMO Testbeds

Design and Deployment Considerations for High Performance MIMO Testbeds Design and Deployment Considerations for High Performance MIMO Testbeds Konstantinos Pelechrinis, Ioannis Broustis, Theodoros Salonidis, Srikanth V. Krisnamurthy, Prasant Mohapatra - University of California,

More information

There are 10 questions in total. Please write your SID on each page.

There are 10 questions in total. Please write your SID on each page. Name: SID: Department of EECS - University of California at Berkeley EECS122 - Introduction to Communication Networks - Spring 2005 to the Final: 5/20/2005 There are 10 questions in total. Please write

More information

Abstract. Testing Parameters. Introduction. Hardware Platform. Native System

Abstract. Testing Parameters. Introduction. Hardware Platform. Native System Abstract In this paper, we address the latency issue in RT- XEN virtual machines that are available in Xen 4.5. Despite the advantages of applying virtualization to systems, the default credit scheduler

More information

Medium Access Control Sublayer Chapter 4

Medium Access Control Sublayer Chapter 4 Medium Access Control Sublayer Chapter 4 Channel Allocation Problem Multiple Access Protocols Ethernet Wireless LANs Broadband Wireless Bluetooth RFID Data Link Layer Switching Revised: August 2011 & February

More information

Implementation and use of Software Defined Radio (SDR) technology for Public Safety, Traffic applications, and Highway Engineering

Implementation and use of Software Defined Radio (SDR) technology for Public Safety, Traffic applications, and Highway Engineering Implementation and use of Software Defined Radio (SDR) technology for Public Safety, Traffic applications, and Highway Engineering Topics of discussion * Section 1. Wireless vs. Wired. Advantages and disadvantages

More information

The Benefits of FPGA-Enabled Instruments in RF and Communications Test. Johan Olsson National Instruments Sweden AB

The Benefits of FPGA-Enabled Instruments in RF and Communications Test. Johan Olsson National Instruments Sweden AB The Benefits of FPGA-Enabled Instruments in RF and Communications Test Johan Olsson National Instruments Sweden AB 1 Agenda Introduction to FPGAs in test New FPGA-enabled test applications FPGA for test

More information

Wireless USB Periodic Transfer Models. Dan Froelich Intel

Wireless USB Periodic Transfer Models. Dan Froelich Intel Wireless USB Periodic Transfer Models Dan Froelich Intel Agenda Wired Isochronous Model Overview Key Features Wireless Media Reliability Coexistence (Shared With Other Hosts And UWB Devices) Wireless USB

More information

SOFTWARE DEFI ED RADIO EXECUTIO LATE CY

SOFTWARE DEFI ED RADIO EXECUTIO LATE CY SOFTWARE DEFI ED RADIO EXECUTIO LATE CY Feng Ge, Alex Young, Terry Brisebois, Qinqin Chen, and Charles W. Bostian Virginia Polytechnic Institute and State University, Wireless @ Virginia Tech, Center for

More information

LabVIEW Communications Application Framework 2.1

LabVIEW Communications Application Framework 2.1 GETTING STARTED GUIDE LabVIEW Communications 802.11 Application Framework 2.1 This document provides basic information about how to get started with the 802.11 Application Framework 2.1. Contents System

More information

Flexible wireless communication architectures

Flexible wireless communication architectures Flexible wireless communication architectures Sridhar Rajagopal Department of Electrical and Computer Engineering Rice University, Houston TX Faculty Candidate Seminar Southern Methodist University April

More information

P B 1-P B ARRIVE ATTEMPT RETRY 2 1-(1-P RF ) 2 1-(1-P RF ) 3 1-(1-P RF ) 4. Figure 1: The state transition diagram for FBR.

P B 1-P B ARRIVE ATTEMPT RETRY 2 1-(1-P RF ) 2 1-(1-P RF ) 3 1-(1-P RF ) 4. Figure 1: The state transition diagram for FBR. 1 Analytical Model In this section, we will propose an analytical model to investigate the MAC delay of FBR. For simplicity, a frame length is normalized as a time unit (slot). 1.1 State Transition of

More information

Multi-Channel Neural Spike Detection and Alignment on GiDEL PROCStar IV 530 FPGA Platform

Multi-Channel Neural Spike Detection and Alignment on GiDEL PROCStar IV 530 FPGA Platform UNIVERSITY OF CALIFORNIA, LOS ANGELES Multi-Channel Neural Spike Detection and Alignment on GiDEL PROCStar IV 530 FPGA Platform Aria Sarraf (SID: 604362886) 12/8/2014 Abstract In this report I present

More information

CM5000 DATASHEET v0.1

CM5000 DATASHEET v0.1 CM5000 DATASHEET - 2 - http://www.advanticsys.com/cm5000.html v0.1 Table of Contents 1. INTRODUCTION... 5 2. HARDWARE CHARACTERISTICS... 6 2.1 CM5000 DIAGRAMS... 6 2.2 MICROCONTROLLER DESCRIPTION - TI

More information

CS263: Wireless Communications and Sensor Networks

CS263: Wireless Communications and Sensor Networks CS263: Wireless Communications and Sensor Networks Matt Welsh Lecture 5: The 802.11 Standard October 7, 2004 2004 Matt Welsh Harvard University 1 All about 802.11 Today's Lecture CSMA/CD MAC and DCF WEP

More information

WP-PD Wirepas Mesh Overview

WP-PD Wirepas Mesh Overview WP-PD-123 - Wirepas Mesh Overview Product Description Version: v1.0a Wirepas Mesh is a de-centralized radio communications protocol for devices. The Wirepas Mesh protocol software can be used in any device,

More information

Computer Communication III

Computer Communication III Computer Communication III Wireless Media Access IEEE 802.11 Wireless LAN Advantages of Wireless LANs Using the license free ISM band at 2.4 GHz no complicated or expensive licenses necessary very cost

More information

Overview : Computer Networking. Spectrum Use Comments. Spectrum Allocation in US Link layer challenges and WiFi WiFi

Overview : Computer Networking. Spectrum Use Comments. Spectrum Allocation in US Link layer challenges and WiFi WiFi Overview 15-441 15-441: Computer Networking 15-641 Lecture 21: Wireless Justine Sherry Peter Steenkiste Fall 2017 www.cs.cmu.edu/~prs/15-441-f17 Link layer challenges and WiFi WiFi Basic WiFi design Some

More information

Improving IEEE Power Saving Mechanism

Improving IEEE Power Saving Mechanism 1 Improving IEEE 82.11 Power Saving Mechanism Eun-Sun Jung 1 and Nitin H. Vaidya 2 1 Dept. of Computer Science, Texas A&M University, College Station, TX 77843, USA Email: esjung@cs.tamu.edu 2 Dept. of

More information

Lecture 6. Data Link Layer (cont d) Data Link Layer 1-1

Lecture 6. Data Link Layer (cont d) Data Link Layer 1-1 Lecture 6 Data Link Layer (cont d) Data Link Layer 1-1 Agenda Continue the Data Link Layer Multiple Access Links and Protocols Addressing Data Link Layer 1-2 Multiple Access Links and Protocols Two types

More information

Midterm Exam CSCE 232: Computer Networks Fall Instructions:

Midterm Exam CSCE 232: Computer Networks Fall Instructions: Midterm Exam CSCE 232: Computer Networks Fall 2007 Last Name: First Name: Student ID: Instructions: 1. This is a close-book and close-notes exam. 2. There are seven questions in total. The number of points

More information

Intelligent Transportation Systems. Medium Access Control. Prof. Dr. Thomas Strang

Intelligent Transportation Systems. Medium Access Control. Prof. Dr. Thomas Strang Intelligent Transportation Systems Medium Access Control Prof. Dr. Thomas Strang Recap: Wireless Interconnections Networking types + Scalability + Range Delay Individuality Broadcast o Scalability o Range

More information

A Multi-channel MAC Protocol for Ad Hoc Wireless Networks

A Multi-channel MAC Protocol for Ad Hoc Wireless Networks A Multi-channel MAC Protocol for Ad Hoc Wireless Networks Jungmin So Dept. of Computer Science, and Coordinated Science Laboratory University of Illinois at Urbana-Champaign Email: jso1@uiuc.edu Nitin

More information

Computer Networks (Fall 2011) Homework 2

Computer Networks (Fall 2011) Homework 2 5-744 Computer Networks (Fall 20) Homework 2 Name: Due: Oct. 2th, 20, 3:00PM (in class) Andrew ID: October 2, 20 A Short Questions. Which of the following is true about modern high-speed routers? A. A

More information

November 1998 doc.: IEEE /378 IEEE P Wireless LANs Extension of Bluetooth and Direct Sequence Interference Model.

November 1998 doc.: IEEE /378 IEEE P Wireless LANs Extension of Bluetooth and Direct Sequence Interference Model. IEEE P802.11 Wireless LANs Extension of Bluetooth and 802.11 Direct Sequence Interference Model Date: November 11, 1998 Author: Jim Zyren Harris Semiconductor Melbourne, FL, USA Phone: (407)729-4177 Fax:

More information

Implementation of Software-based EPON-OLT and Performance Evaluation

Implementation of Software-based EPON-OLT and Performance Evaluation This article has been accepted and published on J-STAGE in advance of copyediting. Content is final as presented. IEICE Communications Express, Vol.1, 1 6 Implementation of Software-based EPON-OLT and

More information

CSE/EE 461 Wireless and Contention-Free Protocols

CSE/EE 461 Wireless and Contention-Free Protocols CSE/EE 461 Wireless and Contention-Free Protocols Last Time The multi-access problem Medium Access Control (MAC) sublayer Random access protocols: Aloha CSMA variants Classic Ethernet (CSMA/CD) Application

More information

Cross Layer QoS Provisioning in Home Networks

Cross Layer QoS Provisioning in Home Networks Cross Layer QoS Provisioning in Home Networks Jiayuan Wang, Lukasz Brewka, Sarah Ruepp, Lars Dittmann Technical University of Denmark E-mail: jwan@fotonik.dtu.dk Abstract This paper introduces an innovative

More information

TCP Over SoNIC. Xuke Fang Cornell University. XingXiang Lao Cornell University

TCP Over SoNIC. Xuke Fang Cornell University. XingXiang Lao Cornell University TCP Over SoNIC Xuke Fang Cornell University XingXiang Lao Cornell University ABSTRACT SoNIC [1], a Software-defined Network Interface Card, which provides the access to the physical layer and data link

More information

CSE 461: Multiple Access Networks. This Lecture

CSE 461: Multiple Access Networks. This Lecture CSE 461: Multiple Access Networks This Lecture Key Focus: How do multiple parties share a wire? This is the Medium Access Control (MAC) portion of the Link Layer Randomized access protocols: 1. Aloha 2.

More information