Routing in cognitive radio networks using adaptive full-duplex communications over IoT environment

Abstract

Wireless communication demands rapidly increase due to the increase of Internet of Things applications, which leads researchers to build secondary networks that exploit the spectrum holes of primary networks. Cognitive radio (CR) technology is adopted in ad-hoc networks (AHNs) rather than infrastructure-based networks because AHNs have lower cost, higher coverage, and easier maintenance compared to infrastructure-based networks. Moreover, in-band-full-duplex (IBFD) in CR networks (CRNs) is gaining the interest of researchers. This mix between IBFD and CRNs brings a great enhancement in the network’s performance due to efficient dynamic spectrum access. Therefore, we propose an adaptive FD-CRNs routing protocol that uses a common control channel. Adaptive FD communication is conducted in our protocol where the secondary users adapt their communication mode based on the primary users’ activity on the spectrum. Communication modes used in our work are FD transmit and sense, FD transmit and receive, and sensing only. The performance of our protocol was evaluated using a java language simulator for IBFD-CRNs introduced previously in the literature. Also, we compare the performance of our protocol with three previous protocols, probabilistic and deterministic path selection in cognitive radio network, and multi-cast half-duplex routing protocol and broadcast full-duplex routing protocol. The performance metrics used are throughput and total execution time.

Appendix

In this appendix, we provide an example where we find the modes of a few randomly positioned nodes in a network designed for illustration purposes. We assume that the source node has data to transmit to the destination node. Obviously, route discovery is required in this case. We further suppose having 4 channels and 11 nodes (including the source and destination nodes). Table 11 shows the availability of the channels at each node. Plus, Table 12 shows information about each of these channels at each node. Interestingly, the information in Table 12 is randomly set, but by keeping in mind the availability set in Table 11.

Table 11 Availability of the channels

Table 12 Information of channels at each node

To begin, the settling stage allows the nodes to see each other. Then, the source node has to start creating its own RRQP before transmitting. Noticeably, right before the beginning of its transfer, the source senses all the channels, detects if the channel is available, not available, or probably available. This indicates the source node about the mode that can be used for each channel. After deciding the suitable mode for every channel, the throughput will be calculated separately. Then, the RRQP gets transmitted to the neighbors.

Node A will be the first neighbor of the source. Of course, Node A starts creating its RRQP, only after receiving RRQP from the source. It finds the suitable mode for each channel then calculates the throughput for each channel. It is noteworthy to know that the throughput for each intermediate node is calculated based on the previously received values from other upstream (towards the source) nodes. Node B and all the intermediate nodes do the same process of RRQP creation after receiving from previous upstream nodes.

Observing that we used random values in Table 12 that go along with Table 11, but the idle probability of each channel has to be calculated using Eq. 4. The detailed RRQP creation process is illustrated in the following text.

Beginning with the source, it is going to find its own suitable mode for every channel in its surroundings. Referring to Eq. 5 and Table 12, the source examines the state of each channel and decides the suitable mode based on the idle probability which was calculated using Eqs. 3 and 4. Table 13 shows the mode decision of the source for each channel.

Table 13 Source decision

This specific procedure is going to be used for each node to find the suitable mode at every channel. Whenever the source finds the favorable mode for its sensed channels, it is going to calculate the downstream transmission time ($T_{{T_{x} }}^{d}$) for each available or probabilistically available one using the Eqs. 6 and 7. As clarified in Table 14, the source calculates the downstream quality/transmission time for $Ch_{1}$, $Ch_{2}$ and $Ch_{4}$. However, $Ch_{3}$ is not available. Besides to, referring to Table 11, $Ch_{2}$ and $Ch_{4}$ are probabilistically available channels.

Table 14 Source's channels qualities

Interestingly, Node A will continue the process after receiving from the source. Node A is going to find the appropriate mode for each channel. Afterwards, it is going to start building its processing message as explained earlier using Eqs. 8 and 9. Continuing our scenario, in Table 15, Node A calculates its processing message’s quality/transmission time values.

Table 15 Processing message qualities for node A

Afterwards, Node A is going to build its RRQP. Referring to Table 11, channel 1 and channel 2 are available at Node A. However channel 3 and channel 4 are not available. In our example, Node A studies $Ch_{1}$ matched with $Ch_{1}$ (itself) using Eq. 10. In other words, it examines using $Ch_{1}$ in two consequent transmissions without switching to another channel. Likewise, it examines $Ch_{1}$ matched with $Ch_{2}$ considering the time required to switch from $Ch_{1}$ to $Ch_{2}$ using Eq. 11. The calculations of the downstream quality for channel 1 at Node A is illustrated in the following table:

Based on Eq. 13, the maximum value of the downstream quality (calculated in Table 16) is 0.65 which is the value extracted when $Ch_{1}$ was matched with itself. Finally, the maximum quality of $Ch_{1}$(which is 0.65) is being compared to the initial value (considered to be − 1) as shown in Eq. 14. Thus, the final value of the downstream quality of channel 1 is 0.65.

Now, the same goes with $Ch_{2}$. Node A examines channel 2 matching with all the other channels in its surroundings and compares the quality/transmission time with the value stored in the processing message and considers the minimum value (Table 17).

Table 16 Downstream quality for Channel 1 (matched with Channel 1 and Channel 2) at node A

Table 17 Downstream quality for Channel 2 (matched with Channel 1 and Channel 2) at node A

After examining $Ch_{2}$ matched with all other available channels ($Ch_{2}$ matched with $Ch_{1}$, and $Ch_{2}$ matched with $Ch_{2}$), the best matched will be chosen based on Eq. 13 and compared to the previously initiated value in the exchangeable message using Eq. 14. Thus, the final value of the downstream quality of channel 2 is 0.399.

The final RRQP for Node A (based on the calculations in our example) is provided in Table 18.

Table 18 RRQP of node A

The same procedure is followed by all nodes in the forward lane and each node builds its own RRQP and broadcasts it to the neighbors until reaching the destination as Figs. 16, 17, and 18.

Fig. 16

Illustration of the first RRQP broadcast

Fig. 17

Illustration of the second RRQP broadcast

Fig. 18

Illustration of the last RRQP broadcast

It is worth noting that Nodes A and B are the source’s neighbors. Thus, the source will broadcast its RRQP to them using directional signals in a (presumed clockwise) “sweeping” transmission pattern over a CCC. After receiving from the source, Node A and Node B will transmit their RRQP using sweeping transmission while receiving from each other as shown in Fig. 16. Note that Nodes A and B are active at the same time, this is because we divide the decisions and transmissions based on stages of neighbors, (like sound waves around the source).

This RRQP broadcast via sweeping transmission continues until reaching the destination as Fig. 18.

In practice, the destination receives information about all the paths established from the source to it. This information includes the accumulative summation of the best modes’ values calculated by each node for the best-selected channel. Then, the path with the highest summation of modes’ values in the forward lane will be chosen as the best path in the back-track lane.

In our scenario, the destination receives information/RRQPs from Nodes I, G and F. Let’s say that the path (Destination, I, H, E, B, Source) has the highest summation of the best modes’ values. The destination will create its “processing” message/table and find the channel with the best mode value between it and Node I. To make things subtler, we decided, in our simulations, that if two channels have the same mode then the channel with the best throughput is chosen in the back-track lane. Afterwards, the destination will start forwarding the chosen channel’s entity/information from its “processing” message as an RREP to the upstream Node I. Directly, Node I will create its RREP then, for comparison reasons with [3], it calculates the throughput for the selected channel and forwards it to Node H. This process will continue in the reverse path until reaching the source to inform it that this path will be used for data transmission.