Due to interference between co-located wireless networks, obtaining accurate channel assessment becomes increasingly important for wireless network configuration. This information is used, among others, for cognitive radio solutions and for intelligent channel selection in wireless networks. Solutions such as spectrum analyzers are capable of scanning a wide spectrum range, but are not dedicated for channel occupation assessment because they are extremely costly and not able to perform continuous recording for a time period longer than a few seconds. On the other hand, low-cost solutions lack the flexibility and required performance in terms of configuration and sensing efficiency. To remedy the situation, this paper presents an alternative for channel assessment on top of a commercial software-defined radio platform. Although there exist software solutions for performing spectrum sensing on such platforms, to the best of our knowledge, continuous spectrum sensing and long-term recording remain challenging. We propose a pioneering solution that is capable of seamless spectrum sensing over a wide spectrum band and guarantees sufficient flexibility in terms of configurations. The proposed solution is validated experimentally. We demonstrate two advantages of seamless spectrum sensing: the capability of accurate channel occupancy measurement and detecting transient signals such as Bluetooth.
In summary, to achieve an advanced wireless system, we need sensing engines with relatively low cost and that are capable of continuous sensing and recording. To this end, this paper presents a solution that is built upon a commercial software-defined radio (SDR)[7]. The solution is further extended on multiple SDR devices for cooperative and distributed spectrum sensing. While the developed solution has less functionality than spectrum analyzers, it is also much cheaper and dedicated for channel assessment. Above all, in contrast to most spectrum analyzers, our solution is capable of continuous sampling and recording.
Spectrum Analyzer pro Live 3.5.11
This section presents some of the most representative sensing solutions, starting from powerful but expensive spectrum analyzers to simple off-the-shelf sensor devices. The processing mechanism of high-end spectrum analyzers is discussed in depth, as it is needed for the further sections.
Some spectrum analyzers combine the swept mode and the FFT mode. The result of one sweep is then a combination of several FFT shots obtained at different center frequencies. This is termed as the swept FFT mode.
The advantage of the FFT-based analyzer is that it is possible to look at a broader range of the spectrum with one operation. However, FFT requires first the acquisition of a batch of samples and then followed by a processing step. What happens in between subsequent acquisition phases is missed by the analyzer, as illustrated in the upper part of Figure2.
FFT-based spectrum analyzer. In the upper part, the FFT processing time is longer than the sampling time, resulting in discontinuous sampling and missed transient signal; in the lower part, the analyzer is capable of detecting the transient signal, thanks to increased processing speed and continuous capturing. The figure is adapted from[8].
The above requirements are also described in the lower part of Figure2. Spectrum analyzers that are capable of seamless measurements are referred to as real-time spectrum analyzers[8]. The exact features of real-time analyzers depend on the type of the device. To illustrate this, two specific spectrum analyzers are described in more detail in the following section: the FSVR series of ROHDE & SCHWARZ and the RSA series of Tektronix. Although modern spectrum analyzers also include the swept mode to increase the frequency range, here we only focus on the FFT mode. The discussion below does not include implementation details but instead emphasizes the underlying processing mechanism and the amount of flexibility for the end users.
Unlike FSVR, the RSA spectrum analyzer from Tektronix exposes more parameters to the end user. For instance, there is a parameter to specify the length of the time interval during which samples are collected and analyzed seamlessly. Both FFT size and sample rate can be configured independently. This decouples the resolution bandwidth from the frequency span. Similar to FSVR, the final result is trimmed by various detectors for displaying on the screen.
Regardless the difference in processing style, the two analyzers do have one thing in common - the output is produced in a way that is best suited for displaying on the screen. Also, in contrast with the fancy display features, the recording features of spectrum analyzers are relatively basic. Both FSVR and RSA are capable of recording raw samples and some amount of spectrogram, depending on the waveform memory depth. Take the FSVR as an example, the waveform memory allows the user to store maximum 200 million in-phase and quadrature-phase (IQ) samples. This means a recording of 8 s with 25 mega samples per second (Msps) of sample rate, which is just wide enough to cover one Wi-Fi channel. Apart from the time limitation, further processing on the raw IQ samples is still required to obtain the energy in specific channels.
In summary, spectrum analyzers are made for fast visualization of various signals and performing sophisticated off-line analysis. The capability of spectrum analyzers are far beyond FFT or storing raw samples; however, they lack the capabilities of continuous recording and fast data transfer, hence are not suitable for basic channel assessment.
The radio of Airmagnet has a 20-MHz intermediate frequency (IF) bandwidth. The Airmagnet spectrum analyzer makes use of the swept FFT mechanism to cover a bandwidth that is wider than 20 MHz. For the 2.4-GHz industrial, scientific, and medical (ISM) band, the entire frequency span is fixed to 83 MHz with 156-kHz resolution bandwidth. During each sweep, the radio increases its center frequency with a step of 20 MHz and dwells on each center frequency for 30 ms. The sweep time is fixed to 1 s. The fact that the frequency span is less than five times the IF bandwidth infers that each measurement of the 2.4-GHz ISM band contains maximum five blocks of samples. This means that the time to sample the wireless medium for each sweep is at maximum 150 ms. Given the constant sweep time of 1 s, the actual sensing efficiency is only 15%.
The mechanism of Wi-Spy resembles the pure swept spectrum analyzer. It uses a narrow-band RF receiver to scan across the interested band in tiny steps. The step width depends on the Wi-Spy model and the selected band of interest, ranging from about 50 to over 600 kHz. Compared to spectrum analyzers, the front-end of USB devices is less advanced, resulting in lower sensitivity and narrower spectrum coverage. Besides the less advanced RF front-end, USB-based sensing solutions rely on host machine softwares for processing, which are typically bound to certain operating systems. This further limits their usage. The feature of long-term recording is provided but with very limited efficiency and flexibilities.
Another option is to use cheap sensor devices for spectrum analysis. Here the sensor devices refer to the battery-operated, low-power wireless platforms[11, 12]. Sensor chips are originally meant to form sensor networks for home automation or various monitoring purposes. They usually consist of integrated sensors, a microcontroller, and an IEEE 802.15.4 (Zigbee)-compliant radio module. The radio module provides built-in clear channel assessment (CCA), which can be used to evaluate the energy of the selected channel. With appropriate firmware, cheap sensor devices can also be programed into a swept spectrum analyzer.
Table1 gives an overview of the advantages and disadvantages of different sensing solutions. High-cost solutions such as spectrum analyzers are usually overkill for the targeted applications: many functionalities built in spectrum analyzers are redundant for merely channel assessment. In addition, these devices are not capable of long-term and seamless recording and fast data transfer. On the other hand, low-cost devices have less bandwidth and limited processing power and flexibility. To remedy this situation, we aim to design a low-cost solution that is capable of seamless recording and offers sufficient flexibility, as listed as the last entry of Table1.
The sensing engine software is currently compiled and tested on six identical hexa-core Linux servers. The choice of using a hexa-core server does not strictly comply to the initial low cost requirement; however, it is necessary to achieve sufficient amount of parallel processing, as further explained in Section 3.3. The architecture of the software can easily be ported to an FPGA platform. Note that the price of one USRP and one server is still significantly lower than the price of one spectrum analyzer; therefore, even the current approach is already an improvement in terms of overall financial cost.
For the continuous FFT mode, the USRP front-end stays at the same frequency and continuously samples the wireless medium. Similar to spectrum analyzers, users can control both the center frequency and the sample rate in order to define the spectrum range.
Table2 demonstrates the performance advantages of our solution. The continuous FFT mode of the USRP sensing engine is the only solution which is capable of 100% sensing efficiency and long-term recording at the same time. The swept FFT mode offers 89% sensing efficiency under a configuration similar to that of Airmagnet, which only provides 15% sensing efficiency. Although the sweep time of our solution is longer than the best record of the (more expensive) FSVR spectrum analyzer, it is still much faster than Airmagnet and Wi-Spy devices. 2ff7e9595c
Comments