Overall Algorithm

With sensing modes and techniques described above, an adaptive algorithm is used to quickly and efficiently determine the availability of a spectrum. A sensing algorithm that benefits from the distributed sensing is described in Figure 4-20.

1- Identify Freq. “0.5” or above 2- Initialize PLL

Perform coarse sweep on freq. with

“0.5” or above Perform RF sweep

was energy detected?

A

1- Update freq. status as a “0.75” 2- Initialize Fine sensing N Y 1- Initialize PLL 2- Perform RF sweep is Freq. “prohibited? ”

1- Update freq. status in LUT as a “0” 2- go to next freq.

was energy detected?

1- Update freq. status in LUT as a “0.25” 2- go to next freq. 1- Update freq. status as a

“0.5” if status is NA 2- Initialize Coarse sensing

A N N Y Y Initialize sensing Enter User Preference and Geo

location Identify candidate frequencies based previous data 1- Identify Freq. “0.75” or above 2- Initialize PLL Perform coarse sweep on freq. with

“0.75” or above Perform RF sweep

was energy detected?

A

1- Update freq. status as a “1” 2- Initialize CR operation N Y Initialize sensing STAGE 1 STAGE 2 STAGE 3 STAGE 4 1- Identify Freq. “0.5” or above 2- Initialize PLL Perform coarse sweep on freq. with

“0.5” or above Perform RF sweep

was energy detected?

A

1- Update freq. status as a “0.75” 2- Initialize Fine sensing N Y 1- Initialize PLL 2- Perform RF sweep is Freq. “prohibited? ”

1- Update freq. status in LUT as a “0” 2- go to next freq.

was energy detected?

1- Update freq. status in LUT as a “0.25” 2- go to next freq. 1- Update freq. status as a

“0.5” if status is NA 2- Initialize Coarse sensing

A N N Y Y 1- Initialize PLL 2- Perform RF sweep is Freq. “prohibited? ”

1- Update freq. status in LUT as a “0” 2- go to next freq.

was energy detected?

1- Update freq. status in LUT as a “0.25” 2- go to next freq. 1- Update freq. status as a

“0.5” if status is NA 2- Initialize Coarse sensing

A N N Y Y Initialize sensing Enter User Preference and Geo

location Identify candidate frequencies based previous data Initialize sensing Enter User Preference and Geo

location Identify candidate frequencies based previous data 1- Identify Freq. “0.75” or above 2- Initialize PLL Perform coarse sweep on freq. with

“0.75” or above Perform RF sweep

was energy detected?

A

1- Update freq. status as a “1” 2- Initialize CR operation N Y Initialize sensing STAGE 1 STAGE 2 STAGE 3 STAGE 4 Figure 4-20 DSR Algorithm

At the core of the algorithm is the ability to continuously update a select number of candidate frequencies for CR operation and learn over time the best bands for CR operation. The algorithm adds a merit factor to each frequency and the ability to continuously monitor the presence of an interferer. Continuous monitoring is done via RF detectors in the analog domain because of its fast response time. As described in previous sections, the disadvantage of RF detectors may be its undesired ability to detect broadband signals which may falsely record energy outside the desired band of interest. This issue can be mitigated by adding filtering. However, the addition of filtering is counter productive, unpractical, and costly. In our implementation, we focus on recording relative energy or simply changes in energy in the band. Relative energy for each band is recorded and saved along with other parameters in the LUT.

As each frequency band is scanned, a figure of merit is attached to it and continuously updated. The DSR assigns 0, 0.25, 0.5, 0.75 or a 1 weight. A “0” associated to a frequency means that the CR is prohibited of operating in the band, while a “1” defines a frequency band at which the CR is cleared to operate. As shown in Figure 4-20, the figure of merit increases as the DSR moves from stage 1 through stage 4. At each stage, the resolution of the frequency detection increases as the sensing bandwidth decreases. The ability of the algorithm to quickly discard large frequency band as non-candidates via the initial setup or RF scanning is important since the CR does not waste time and power on known non-candidates. Obviously, this algorithm can be easily augmented with additional levels of sensing. As an example, if a cyclostationary feature sensing is added to the sensing receiver prior to initiation of CR operation, a “0.85” figure of merit may be added to the sensing weights.

The algorithm above takes full advantage of the combined benefits of sensing at the different stages of the receiver. In addition, it is able to localize some of the decision making close to the sensing point. The advantages of this technique are:

1- The proposed approach takes advantage of the availability of two antennas and two receivers (Main and DSR) in essence providing the benefit of diversity during sensing. 2- Better decision making: with a dedicated channel, the sensing receiver continuously

senses and learns without the added complexity or urgency to send and receive data packets. The sensing receiver scans the desired frequency bands and records the frequency

bands where a CR is allowed to operate. Assuming the CR may coexist with a PU, the receiver learns and generates a prediction algorithm of the probability of PU appearance or interference with other CRs.

3- Faster scan: the sensing receiver prioritizes and saves the list of “best” channels in the look up table. In addition, the look up table may save the “preferred channels” where the CR has been historically operating. Hence, upon power up of the radio at stage 1, the main receiver searches the “preferred channels first” prior to doing a broadband sweep.

4- Built-in diversity leverages the availability of two receiver (main radio and DSR) chains and two antennas. Diversity is an effective tool to combat fading and multipath in the wireless channel. Although the DSR is dedicated to sensing at all times, the main receiver assists the DSR in 1) coarse sensing after the initial scan, and 2) fine sensing once an initial list of candidate frequency bands are identified for CR operation or during “quite” time.

In order to take full advantage of the DSR and its algorithm, the radio architecture and especially the phase locked loop must be able to quickly hop and settle onto a desired frequency. Without an agile PLL, the system scan time would be gated by the radio hardware. The overall PLL design is critical to the performance, cost, and complexity of the CR specifically across wideband operation.