Researchers believe that one of the causes for the recent increase in carbon dioxide and serious environmental problems worldwide is the rise in the amount of data due to the advancements in information and communications technology. Yet, the effort to reduce the quantity of CO
2 or data alone has not shown enough effect. Thus, a more sound communication technology should be introduced to resolve the problem. The energy harvesting capable cognitive radio (CR) network, which senses its environment and surroundings, is effective in the development of eco-friendly, low-carbon protocols since its nodes can each recognize the energy consumptions and environmental changes [
1]. Previous studies on the CR network have focused on achieving high efficiency in radio frequency. However, considering the fact that the cognitive network has three stages in its process sensing the environment, deciding the optimum action, provided that a device with cognitive function can accurately sense and conserve the available frequencies and energy consumptions of different operations, it can well be applied to green, low-carbon communication protocols. Network coding reduces the amount of data and unnecessary data retransmissions and provides the estimated transmission patterns of network nodes. This technique can cut down the consumption rate of wasted energy, thereby contributing to the enhancement of the system performance. In a linear radio network-based CR network, the patterns of data transmission and reception of a primary user (PU) can be assumed by the continuity of
k linear equations, and thus a secondary user (SU) can predict the duration of PU’s frequency use in a huge probability [
2,
3]. As a result, in a CR network that utilizes the linear network coding technique, while the transmitted and received data of the PU is well protected, the SU can make the maximum use of vacant frequencies without wasting energy through unneeded sensing, maximizing the system performance. It is still impossible to detect PUs perfectly, so there are cases of false positive errors caused by wrong judgement of the nonexistent PU as existing and false negative errors caused by vice versa. Analog network coding is a technology that uses the interference inversely [
4]. In this technique, two different nodes intentionally transmit wireless signals at the same time, and the middle node receives one integrated signal. Each source node and the nodes within the overhearing range possess the original data, so when the middle node broadcasts the integrated packet, each node destination can decode the wanted packet. This coding allows the packet to be transmitted one time less compared with the existing digital network coding [
5], so it can cut the total number of packets into half compared with the uni-cast transmission method which does not use network coding technology at all. In a study by Katti et al. [
4] to generate the chance for active noise control (ANC) deliberately, the middle node had to hint (trigger) the source nodes to transmit almost simultaneously with a short delay. The reason for such process is that to integrate ANC signals, two different signals should be transmitted at once, but at the same time, the preamble of each signal should not be interfered for decoding. In addition, the previous study had not been aware of the energy harvesting and focused only on a few particular topologies to gain profits, such as Alice-Bob topology which uses the same frequency and X-topology which has a perfect overhearing range [
2]. Moreover, even when two different original data were mixed on the physical layer of the middle node, one of the data should be broadcasted again to each node destination to decode the other one using changed amplitudes and phases of the wireless signal. In this paper, we suggest a technique of restoring a naturally integrated signal of PU and SU with ANC technique, using the minimum amount of energy from the CR network, which functions network coding and energy harvesting. In false negatives, PU receives some of
k as a linear network coded packet and the other as a mixed signal with SU packet. Even if PU’s signal interferes with SU’s signal, PU can still decode the packet he or she wants because the only difference the error brings is the coefficient of linear network coding packets. In false positive, SU can benefit because it can successfully transmit the packet by its own or harvest energy. However, unlike existing studies on CR network, we propose a technique for eco-friendly, low-carbonate communication by which SU can predict and recognize the frequency, the energy consumption of PU and the gain of energy harvesting, thereby choosing the optimum in false negatives or false positives. This process is modeled by partially observable Markov decision process (POMDP). In the “
Related work” section, we introduce related work. In the“
Stochastic policy based wireless energy harvesting in green cognitive radio network” section, we illustrate the eco-friendly, low carbonate, energy harvesting capable CR network technique which this paper proposes. Then, in the “
Performance evaluation” section, we evaluate the performance. Finally, “
Conclusions” section concludes the paper.