Dynamic Spectrum Access Decisions. George F. Elmasry
Читать онлайн книгу.
steps towards grasping the many facets of spectrum sensing through covering the foundations of many of the known techniques to help the system designer select the most appropriate sensing technique for a given system.
Figure 2.1 illustrates how sensing a wideband of frequency can be expressed as spectrum usage in terms of frequency and amplitude. A spectrum sensor can divide this wideband into small sub‐bands, with each sub‐band defined by a frequency range where the spectrum sensor performs sensing over all of the sub‐bands in parallel. An important objective of spectrum sensing is the ability to sense a wideband of frequency divided into sub‐bands so that the information output of the spectrum sensor can help the DSA techniques find different spectrum opportunities.
Figure 2.1 Sensing a wide band of spectrum.
Notice how the vertical access unit in Figure 2.1 is dBm, which stands for decibels (dB) below one milliwatt (mw). This unit is commonly used for energy detection which uses the one milliwatt as a reference and bridges the gap between SNIR, which is in dB,2 and energy detection in watts. With this normalization, a signal received at 1 mw maps to 0 dBm while a signal received at 0.1 mw maps to –10 dBm. Energy detection can be expressed as received signal strength indication (RSSI), which uses the unit dBm. This conversion uses the formula dBm = 10 × log(P/1 mw), where P is signal power.
2.2 Time, Frequency, and Power Spectrum Sensing
Although energy detection is the most common spectrum sensing technique, time, frequency, and power spectrum sensing are covered first in this chapter to emphasize the multidimensional aspects of spectrum sensing. The idea behind this spectrum sensing technique is to create multidimensional spectrum awareness. The simplest form of this spectrum sensing technique is a two‐dimensional spectrum sensing that uses the frequency and time dimensions. Figure 2.2 shows this two‐dimensional spectrum sensing where the spectrum sensor looks for occupancy of certain frequency bands at certain times.
Figure 2.2 Two‐dimensional spectrum sensing.
A secondary user can use this spectrum sensing technique where it can hop to a different frequency band once it detects another user on a frequency it is using. This technique does not consider the signal power and relies on DSA defining a cutoff RSSI level to consider a frequency band as occupied or can be opportunistically used. The cutoff RSSI level may also be an estimation of additive white Gaussian noise (AWGN) without the presence of any signal.
If a communications system is to consider underlay or overlay transmission, the two‐dimensional spectrum sensing in Figure 2.2 can be turned into three‐dimensional spectrum sensing as shown in Figure 2.3. In this case, the power dimension is added. In Figure 2.3, the unoccupied areas are often referred to as spectrum holes or white spaces.
Figure 2.3 Three‐Dimensional spectrum sensing.
Notice that more dimensions can be added to this multidimensional spectrum sensing technique. For example, the spreading code can be sensed and made a fourth dimension. Spreading code sensing can show opportunistic transmission based on using specific spreading codes. Spectrum sensing of spreading code is covered in Section 2.3.3 as a signal characteristic.
2.3 Energy Detection Sensing
This is the most common spectrum sensing approach used today. As explained in Chapter 3, the receiver's operator characteristic (ROC) function makes good use of this simple energy detection approach. Chapter 3 covers how same‐channel in‐band sensing can use energy detection sensing with minimal requirements on the receiver to hypothesize the presence of interference.
RSSI is a common expression of energy detection. RSSI is so common that you can look at your phone while having wireless fidelity (WiFi)3 connectivity and count the WiFi connectivity indicator lines on the top of your screen to see how RSSI is commonly mapped to about four levels (no connectivity, low connectivity, medium connectivity, and good connectivity). Laptop WiFi connectivity indicators typically map the WiFi RSSI to five or six levels. Other devices illustrate RSSI in more or less this number of levels.
A simple receiver can collect the energy received on the antenna in a certain frequency band and quantize it. Low computational complexity and simple implementation are what makes energy detection commonly used. In Figure 2.3, energy detection below a certain power threshold constitutes an opportunity for a spectrum band use (e.g., by a secondary user4). On the other hand, energy detection above that threshold constitutes an occupied band. In Figure 2.3, energy detection is the power axis. Deciding the value of that cutoff threshold can be challenging as the primary user signal may suffer from interference, multipath fading, and jamming among other factors that can affect the signal strength. Energy detection becomes especially more challenging when sensing spread spectrum signals that tend to have low energy. Chapter 3 is dedicated to DSA decision making and will discuss how cutoff thresholds can be used.
2.3.1 Energy Detection Sensing of a Communications Signal (Same‐channel in‐band Sensing)
Let us start from the unit of energy of a communications signal Φj(t), which can be defined as follows: If the receiver detects a 1 V signal across a 1 Ω resistor, the integration of the square value of signal voltage over a specific time period (Tg, Tf) is 1, that is, the receiver has detected one unit of energy.5 Notice the following:
1 The signal can be constructed in a multidimensional signal‐in‐space (SiS) as a vector.6
2 The time T = Tf − Tg is a critical factor in detecting the signal energy.7 If the signal is too weak, the integration of the square value of the signal voltage may need a long period of time to yield reliable energy detection.
The receiver of a communications signal detects a multicoefficient signal in N‐dimensional SiS and attempts to match the received signal with one of M signals.8 The energy detector cares only for the signal energy not the signal decoding.
The signal's ith dimension projected on the kth base can be expressed as follows:
(2.1)
where Φk(t) is the signal basis per coefficient.
Notice