Challenges Of Implementing Full Duplex Communication In Personal Communication System

Discuss about the Challenges of Implementing Full Duplex Communication in Personal Communication System.

Communication is an important activity by which exchange of thoughts and ideas take place. Point to point communication is referred to as a duplex communication system that is made up of two parties attached or two devices connected to each other, such that communication might take place from both the sides, in both the directions (Hossain and Hasan 2015). The term “Du” comes from the word double and the term “plex” refers to structures. Thus, a duplex communication is defined as transmission of data and communication of information in only one direction at a time (Wang et al. 2016). Duplex communication might be of two types: full duplex communication and half duplex communication. Since with the enhancement of communication system, he communication networks deliver a significant amount of data, the spectral efficiencies of the network needed an improvement (Thilina et al. 2015). Some of the advanced technologies such as the MIMO (Multiple-input-multiple output), as well as the orthogonal frequency division multiplexing were some of the promising solution for enhancing half duplex communication, yet the desired efficiency in communication was not achieved (Zhang et al. 2015). In order to fulfill the shortcomings of the half duplex communication, full duplex communication was introduced.

In theory, full duplex communication the signal being transmitted is known to the sender and the circuit design is only considered to remove it from the signal that is received. It follows that the usual way of abstraction that the analog radio is a black-box which takes input, the digital baseband signal and then changes the signal into a digital one (Zhou et al. 2014). Radios distort the signal that is transmitted and hence the digital baseband representation is also distorted.      

Full duplex communication refers to the communication system where both the parties communicating with each other simultaneously (Zhang et al. 2016). One of the best examples of duplex communication is using a telephone. In a communication over a telephone, a person can speak and hear simultaneously. The earphone in the telephone produces the speech of the listener, whereas the microphone transmits the speech to the listener (Liu et al. 2015). A full duplex communication system allows communication in both the directions simultaneously. The telephone system along with the mobile phones used in the modern days is also examples of full duplex communication (Song et al. 2015). In almost all of the full duplex communication system, the message sent is not perceived to be sent, until it is acknowledged by a response from the receiver. The full duplex communication system uses a channel access method by dividing the forward and reverse communication channels, on the same physical media, through which communication takes place (Liao et al. 2015). The duplexing methods are time-division duplexing and frequency division duplexing.

With the rapid evolution in the wireless communication and the enhancement in the communication techniques, resulted in a huge amount of traffic and the number of wireless devices exceeded the spectral limit (Zhou et al. 2014). The end of the year 2011 witnessed, 86% increase in the  number of mobile phone subscribers than the previous year. With the global growth of the mobile traffic data, the mobile traffic grew to 66%, during the next five years (Liu et al. 2014).   

Evolution of technology and need for full duplex communication

Figure 1: Relative distribution of the global mobile data traffic over different portable UEs

(Source: Liu et al. 2014)

The modern trend in the wireless communication and specifically with the increase in the use of the radio systems, it was important to find a solution to improve the spectral efficiency as well as to increase the flexibility in the spectrum use. The increasing capacity of the mobile communication combined with the slow pace of reallocation of the new spectrum, use of full duplex communication became a necessity. The full duplex communication features could be successfully used to enhance the wireless communication. The features are:

Increase in the link capacity: Full duplex communication doubles the capacity of the link as compared to the traditional half duplex mode.

Enabling introduction of the novel and efficient channel access mechanisms: Wireless devices that are capable of full duplex communication, listens to the signals as well as transmits them. This is done in order to probe if any other transmission occurs in the same radio channels (Zhou et al. 2014).   

Reduction in the air inference delay: Simultaneous receiving of the acknowledgement along with transmission of data enables shorter latency in the data transmission.  

Increase in flexibility in the spectrum usage: The resources with the same frequency could be used for a bi-directional communication. In case of bi-directional communication, the transceiver might be able to communicate with one full duplex communication device (Zhou et al. 2014).  

Novel relay solution: The reuse of the spectral resources enables most instantaneous retransmission.

Improvement in the ad hoc as well as mesh network operation: Full-duplex communication might assist to get rid of the hidden issue that is seen for the mesh networks.   

Improvement security in transmission: Two transmitted signals are mixed in the same carrier. As a result eavesdropping is complicated.  

Time division duplexing (TDD) is a duplxing method in which time division multiplexing is done to separate the outward and return signals. TDD is advantageous where there is asymmetry of the downlink and the uplink data rates (Prendergast et al. 2015). With the increase in the amount of uplink data, more communication capacity is allocated dynamically. For communication systems such as the radio systems that are not quickly moving, the uplink and the downlink channels are similar to each other (Zlatanov et al. 2014). Thus, the technique of beam forming works efficiently in the time division duplexing system. Time division duplexing includes DECT wireless telephone, IEEE 802.16 WiMax.

The frequency division duplexing refers to the process, in which the transmitter and the receiver operate at different carrier frequency. The stations need to be efficient enough to transmit and receive the messages simultaneously (Gallinaro et al. 2014). This is done by altering the frequency of the sender and receiver signals slightly. Thus, duplex communication is carried out successfully, by slight alteration of the frequency of the sender as well as the receiver. The uplink as well as the downlink is separated by frequency offset (Rodriguez et al. 2014). This method is effective in case of symmetric traffic. In case of the TDD, the band-width is wasted during the switch over from the transmitting mode to the receiving mode (Zhou et al. 2014). However, in case of frequency division duplexing, the base stations do not listen each other and hence does not interfere with each other (Chin et al. 2014). However, in TDD, guard timings needs to be kept, to avoid interferences among the neighboring base stations. Examples of FDD include ADSL and VDSL. IEEE 802.16 WiMax is also an example of FDD (Bharadia et al. 2013).

Time division duplexing

Echo cancellation is an important activity in order to ensure effective full duplex communication. In audio duplex communication systems such as telephones create echoes. This needs to be eliminated (Sabharwal et al. 2014). The occurrence of echo takes place when a sound coming from a speaker that originates at a far end reenters the microphone and is returned to the far end (Ju and Zhang 2014). The process of echo cancellation refers to the cancellation of the signal processing that subtracts the far end signals that comes out of the microphone signal, before it is echoed back.

The wireless communication system has been designed with the assumption that time division duplexing and frequency division duplexing is being implemented. The assumption of the half duplex communication has a fundamental effect across the various protocol layers.

At the physical layer, the use of the full duplex is advantageous since it doubles the sum capacity or the spectral efficiency. However, throughput is not the only parameter for quality of service. The full duplex communication has a significant effect on the packet delay as well (Zhou et al. 2014). The capacity of the network layer is also affected by scheduling as well as resource allocation protocols. The signal-to noise and interference ratio is one of the most significant metrics that is affected by the full duplex communication. The reason is the strong coupling of the node’s transmitted signal with the received signal.

At the MAC layer and link layer, the full duplex communication has a potential to reduce the MAC delay as the transceiver starts transmission while receiving and might be able to transmit both data frames as well as signaling message. However, the full duplex transmission degrades the reliability of the link, since the SINR is reduced at the physical layer.

At the network layer, the achievable throughput could be increased since it does not have to put restriction on the transmission and receiving of the packets. The end-to-end delays could be decreased since acknowledgements could be sent at the same time than the packets are received (Liu et al. 2015). The end-to-end delay has a strong effect on the queue lengths of the nodes. The node buffer space requirements are significantly reduced since the node does not need to stop its sending process, for the acceptance of acknowledgement. The delay jitter might be decreased, since delay in queuing could be decreased, as the nodes need not wait to receive the responses. The packet loss ratio could be increased due to the process of self interference (Zhou et al. 2014). The energy expended for each packet could be increased due to the required self interference cancellation. The route lifetime could be decreased if the power consumption of the node is increased due to self-interference cancellation. Full duplex implementation is a complex process and it requires additional hardware cost. Moreover, the full duplex technology might ot be able to be implemented in the first phase (Korpi et al. 2014). However, in case of pico-base station and femto base station, the full duplex communication could be implemented. A base station can receive signals from one of the nodes at a time along with transmitting to another node simultaneously. A base station is able to communicate with four base stations simultaneously, with two uplink and two downlink transmissions (Zhou et al. 2014). If the base station is capable of full duplex communication, it might successfully allocate two radio sources in order to support four transmissions at a time. It is assumed that the LTE small cells use frequency division multiplexing or time division multiplexing, as a multi user access scheme. In these types of the duplex transmission, two kinds of interference dominate the performances. The base station suffers from the self-interference for its own transmission. Two of the nodes communicating with each other suffers from inter-user interference while communicating with the other two nodes. However, the challenge faced is to measure co-channel interference, at various nodes. In order to have the maximum gain, the base station needs to schedule a pair of the transmit nodes and receive nodes that operate on the same source. If the frequency division multiple access scheme is used, then additional challenges might be faced. One of the additional challenges includes recovering from the additional interference that is caused due to leakage of the signal that is being transmitted on the neighboring bands. In the full duplex mode, strength of the signal leakage is much higher due to the additional transmission. Another significant challenge faced in the full duplex communication is caused by the transmission of the nodes that are nearby and which are operating in the adjacent frequency band. Though the neighboring nodes are allocated to the orthogonal frequency band, the transmission power might still saturate the radio frequent chain to the receiver node.

Frequency division duplexing

Figure 2: FDMA multi-user access scheme

(Source: Korpi et al. 2014)

Figure 3: TDMA multi-user access scheme

(Source: Korpi et al. 2014)

In this type of application, there are no chances of co-channel interference. The self interference in both the base stations as well as in the communicating node, dominates the performance. Since the self-interference is not stable, hence the user scheduling and resource allocation might not be a complicated one. Another significant challenge is the traffic load. When the bidirectional traffic is not symmetric, a new mode called the hybrid mode could be applied. However, in the hybrid mode, the co-channel interference problem might arise. The full duplex communication technology could be used by implementing two user scheduling methods. These are the time division multiple access with multi-user access scheme, along with time division multiple access with multi-user access scheme. With this method, the base station allocates a radio source for transmission as well as reception, along with allocating another orthogonal radio source. The device-to-device communication is an added feature into the LTE release. It is considered that two local UEs might communicate with each other on the macro source that is unused, if the adjacent node is not using it. In order to ensure that there is no interference of the macro cell communications, the low transmission power is preferred for device-to-device communication. If only one node is capable of full duplex communication, then the communication among the nodes might be improved by using other resource allocation methods. The IEEE standards specify three main techniques to handle access in a shared medium. These are as follows:

DCF: The distributed coordination function ensures that there is no delay in the random access to the wireless medium.

PCF: The point coordination function uses a method of polling technique that is initiated by the access point such that the access to the shared link is coordinated. This allows quality of services and delay control.  

HCF: The hybrid coordination function is a combination of the previously mentioned techniques.

The distributed coordination function is the most widely spread techniques. This communication technique does not require synchronization between the communicating nodes. This technique could be used with ad-hoc mesh mode. This is based on the carrier sense multiple access (CSMA). The CSMA technique is able to coordinate access to the wireless channels. The DCF also uses a set of timers as well as an exponential backoff, in order to resolve the collision issues (Vermeulen and Pollin 2014). The inter-frames are initiated, before as well as after the transmission, with a silent period exponentially distributed.     

Design protocols differ widely in half duplex and full duplex modes. The different protocol stack layers could be optimized based on the optimization problem formulations. Since in case of full duplex communication, the communication process takes place simultaneously, hence dedicated protocol stack design is essential. The size of the interference region depends on the transmission power of the node. The full duplex operation sets a limit for the transmission power due to the imperfect self-interference cancellation, thus leading to the receiver’s limited self-interference tolerance. This decreases the interference region of a single node, along with the decrease in the range of transmission. The reduction in the transmission range due to the full duplex operation and results in more hop of the ad hoc network. The MAC protocols ensure the efficient and effective sharing of the files using the wireless bandwidth. The currently used protocols use the time division multiplexing and the frequency division multiplexing. However, using this process, collision detection is not possible. Hence, in MAC protocol, the carrier sense multiple access with collision avoidance (CSMA/CA) is used. The full duplex communication allows more number of users to use the communication channels. However, since the uplink and down link transmission occurs at the same time slots, the scheduling for both the traffic becomes hectic. In case of full-duplex communication, the scheduling algorithm could be prevented. In order to overcome the problem of hidden node, the IEEE802.11 protocol implements a virtual carrier sensing techniques. This method is used based on two handshake messages between the receiver and the sender. After listening on the media, the source node starts with sending a request to send (RTS) frame. This is intended to be send to the receiver, before the communication starts. If the receiver receives the RTS successfully, and is available for the time that is defined in the (Network allocation vector) NAV, it send a clear to send CTS frame. This CTS frame is used to inform all the neighboring nodes that the node is about to start sending messages. Only, after the initial handshake process is fulfilled, the data packets are transmitted along with the acknowledgment message.

Since the IEEE 802.11 protocol suffers from poor performance as the number of users increase, implementation of the full duplex communication would enhance the performance of the 802.11 networks (Zhou et al. 2014). The scheduled receiver’s transmission does not affect any other station’s delay. The full duplex communication allows gain of the throughputs. All the exchanges are included inside the RTS-CTS handshake such as signaling protocols.

FDD based full duplex communication access the uplink and downlink. In this kind of configurations, the uplink and the downlink operations take place on different set of frequency bands that are disjoint. Thus, FDD based full duplex communication is an efficient one.

TDD based full duplex communication, the uplink as well as the downlink takes place over the same frequency bands, but using different time slots. In this mode, the terminals might use the downlink slots to transmit to the receiver. This allows receiving and sending at the same time. Gains could be achieved when the uplink time slots of the full-duplex communication receives information from various other nodes (Zhou et al. 2014). With the growth of the number of full duplex communication users, the full duplex communication could be exploited. Thus, an user might be able to observe the expected gains with the increase in number of full duplex communication terminals. However, this gain is bounded by the number of timeslots that are available for communication in the system. However, a minimum number of full duplex communication users are needed for compensating the signal overhead. Moreover, the throughput gain is significantly impacted by the number of full duplex slot used. The gin in terms of the total added capacity in the network is directly proportional to the number of the full duplex timeslots (Zhou et al. 2014). However, the expected gain cannot be accounted successfully, without taking into consideration, the overhead of the new technology. This overhead includes, exchange of the additional messages, dedicated time slots for the purpose of signaling, along with control information.                          

The advantages of using the full duplex communication in personal communication are as follows:

1. Throughput gain:The full duplex communication nearly doubles the throughput in comparison to the half duplex communication.
2. Avoidance of collision:The carrier sense multiple access (CSMA) with collision avoidance requires half duplex nodes to check the quality of the channel before using it (Sabharwal et al. 2014). However, the full duplex mode requires the initial node that starts transmission to sense the channel, before data is transmitted through it. This is done to avoid collision of the packets.
3. Solution to the hidden terminal problem:The issue of the hidden terminal could be solved by using the full duplex communication. A scenario is considered with multiple nodes that has data in their buffer for direct transmission (Hui et al.2016). If any of the nodes starts transmission of data, to the access point, this point also starts data transmission simultaneously, back to the node. In this situation, the other nodes would hear the transmission of the access point and delays the transmission. This delay is done to avoid collision. Even if the access point has no data to be send to the initial node, an acknowledgement is send to the node, thus, preventing the other nodes to transmit at that very moment (Zhou et al. 2014).
4. Reduction in congestion using the help of MAC scheduling:The loss of throughput that is imparted by congestion could be circumvented by using the full duplex communication mode (Liu et al.2015). With the assistance of full duplex operations, the node 0 is able to transmit as well as receive simultaneously. Thus, with the help of full duplex operation, node 0 is able to do both transmitting and accepting all the while, consequently the total system throughput might approach the single-connection limit, while at the same time profiting from the spatial assorted qualities pick up.
5. Reduction of the end-to end delay: A full duplex node is only capable of forwarding packets that are partially received such that the end-to-end delay is reduced significantly. The full duplex communication uses the technique of multiple hop and forwards partially received packets also (Ahmed et al. 2013). However, in case of half duplex mode, the technique of store and forward is used, thus increasing the end-to-end delay.
6. Enhancement of the user’s detection quality in cognitive radio (CR) environment: The detection of a primary user is a difficult task in the cognitive radio environment. This task is even more complex, if the primary receivers operated in the half duplex mode. The advantages presented by the full duplex mode is that, it enables the secondary users to scan in search of primary users. The primary users might operate at the same time such that the secondary users might find it easy to scan and detect presence of any other nodes.  

The main goal of full duplex communication is to receive and transmit messages simultaneously, within the same frequency. However, in case of full duplex communication, the device receives the signal of interest along with signal it is transmitting. This results in the self interference (Zhou et al. 2014). Moreover, it has been observed that the strength of the self interference signal is 50-100 dB, which is much stronger than the signal of interest. The strength of the self interference signal gains control over the settings of AGC, which is used to scale the input signals, before digitalization (Yun 2016). The high SI power, invokes higher quantization of noise on the signal of interest, and eroding the SINR. The main components in the self-interference could be categorized into three classes:  

1. Linear component: This component represents two important tones that are attenuated and consists of reflections from the outer environment.The components are linear since the received distortion could be represented as a combination of original two tones copies that are delayed.   
2. Non-linear component: These are the components that are created since the radio circuits take an input signal and create the output that contains non-linear higher order terms and cubic terms. The signal terms of the higher order have a high frequency that is similar to the frequencies of the transmitted signals (Zhou et al. 2014). The signal terms of the higher frequencies are close to the frequencies that are transmitted. These signals directly correspond to all the harmonics. The harmonics are the distortions that occur at equally spaced intervals of frequencies. Spike frequency is observed at 2.447GHz as well as 2.453 GHz. These are spaced 2 MHz apart from each other.        
3. Transmitter noise: It has been observed that,the increase in the base signal observed is seen on the two main tones that is the noise obtained by the use of the radio transmitter. The noise of the radio is 90dBm. However, as the power flowing through the side bands are much higher, on the levels of 50dBm higher than the noise of the receiver (Cheng et al.2015). This noise that is extra is being generated from the components of high power of the radio transmitter. Moreover, the radios have phase noise that is generated by using the local oscillators, of the level of 40 dBm or 50 dBm.

Figure 4: Self-Interference cancelation and its types

(Source: Cheng et al. 2015)

The analog cancellation circuits as well as the tuning algorithm provide 60 dB of self-interference cancellation. One antenna that is attcahed to port 2 that provides limited isolation. However, the circulator cannot isolate port 1 and port 3. The leaking self-interference from the circuit is reduced by 15 dB only.

The digital cancellation aims towards cleaning of the remaining residual self-inferences signals, after the analog cancellation (Ali et al. 2014). The digital cancellation cancels the linear signals by 50 dB and the non-linear component by 20 dB. The digital cancellation ha two parts: canceling the linear components and cancelling the non linear components.

Figure 5: Self-Interference cancellation

(Source: Cheng et al. 2015)

The fundamental phase of computerized cancelation mitigates the straight segment that is stayed as remaining of the self-impedance. This contains the primary transmitted flag that holes over through the circulator, as the staying of the simple cancelations. The reflections are additionally constricted and deferred by fluctuating sums. Keeping in mind the end goal to wipe out the straight and also non-direct segments, it is required to evaluate the self-impedance at the present moment.

The next phase of digital cancellation is the elimination of the residual non-linear component with signal strength of approximately 20dB. This phase takes place after the reduction of 60dB using the analog cancellation. It is difficult to guess the particular non-linear function that a radio might be applied to the transmitted signal by the baseband. The complexity of the digital cancellation is a great overhead for self-interference cancellation (Ali et al. 2014). The two metrics that needs to be evaluated in context to the digital cancellation are: increase in the noise floor and SRN loss. The residual component of the interference that is left after the cancellation of the self interference manifests itself as the increase of the noise floor from the signal received. If after the cancellation it is observed that signal energy of 88dB, is still remaining, then it could be concluded that the noise floor has been increased by 2dB. The signal to noise ratio (SNR) loss is decreased in the SNR by the signal that is received, when the radio is operating in the full duplex mode. In order to perform this, it is essential to measure the SNR of the signal that is received, while the radio is operating in the half duplex mode, with no self interference (Sabharwal et al. 2014). This process is continued with the radio operating at the full duplex mode. The difference of the SNR is measured as the SNR loss.

In order to implement full duplex communication, in personal communication, various challenges are found. The various barriers and challenges in the implementation of the full duplex communication are as follows.      

The main challenges in the implementation of the full duplex communication are the large difference in the power level between the transmission of the transceiver and the signal of interest that is coming from the distant source. Moreover, powerful compilation of the antenna, baseband solutions along with RF is needed for the successful implementation of the full duplex communication. With these factors being fulfilled, the transceiver can transmit and receive messages simultaneously (Tang and Wang 2015). Thus, full duplex communication could be achieved successfully. However, in practice, the self-interference cancellation is limited, and cannot be fully achieved successfully. Moreover, the achievement of the successful full duplex communication is dependent on multiple factors of the wireless devices.        

One of the main challenges while implementing full duplex communication in personal communication is implementation of the full duplex wireless device with a large difference in power between the self-interference that is imposed by the own transmission of the device and the signal that is received from the remote source (ElSawy et al. 2016). The self-interference cancellation or echo cancellation techniques could be divided into three categories. These are: passive suppression, digital cancellation and analog cancellation (Mahmood et al. 2015). The main impairment and barrier to implementation of full duplex communication are the phase noise, nonlinearity of the amplifier and the in phase and quadrature phase imbalance (Kim and Stark 2014). These are some of the issues that degrade the self interference cancellation.

At the level of the antennas, it involves the use of two transmit antennas and one receiver antenna. Offsetting of the two transmitters by half of the wavelength, results in acting destructively in cancelling one another (Mahmood et al. 2015). This results in creation of a null position, in which the receiver antenna is able to receive the other even though it is a weaker signal. At the level of the analog circuits, the self-interference cancellation in the analog domain uses a noise cancellation circuit. The transmission signal is then send to the circuit as noise references that are subtracted from the received signals. The digital baseband receives digital samples after the analog-to-digital conversion is used in the received path. This transmitted sample is stored in the local memory.

Another downside of full duplex communication is that none of the full duplex techniques could be experimentally attained. This is because, it is impossible to attain the theoretically increasing two times the gain in terms of throughput. The practical platform of the full duplex communication suffers from a signal-to-interference and noise radio (SINR) loss due to the strong and significant impact of self-interferences. The impact of self-interference is caused by huge differences in the power between the interference signals. These interference signals are imposed by the device’s own wireless transmission (Thilina et al. 2015). Moreover, heavy self-interferences might result in reduction in the capacity for the full duplex systems that might fall below the half duplex systems (Korpi et al. 2014).

Various techniques of the self-interferences cancellation has been studied by various researchers. Radio frequency cancellation could be employed to achieve full duplex operations (ElSawy et al. 2016).  Moreover, in order to perform the process of self interference, cancellation and passive suppression is carried out. The main enhancement with the self interference cancellation or suppression in full duplex communication is highlighted as follows:

Before the process of analog or digital domain in self-interference cancellation, various techniques that are capable of achieving a high isolation are applied (Kim et al. 2015). The high isolation is achieved between the transmission antennas and receive antenna could be successfully employed to suppress the self interference strength (Zlatanov et al. 2014). Cancellation of analog is capable of preventing the high power self interferences, that is inflicted by the analog to digital convertor. This would desensitize the automatic gain control that owes to the signal leakages. After performing the analog cancellation, the residual self-interferences that is encountered in the practical systems remains the rate-limiting bottle neck (Goyal et al. 2014). Note that no remain solitary advanced or simple strategy is equipped for getting a high cancelation capacity that fulfills the deciphering prerequisite. The obstruction dismissal proportion of the simple strategies ranges from 20 to 45 dB. Thus, it has been watched that mix of the simple and advanced cancelations may be adequate to offer a high self-impedance cancelation ability (Chung et al. 2015).

Another challenge faced by implementation of the full duplex system in personal communication is non-linear distortion and non-ideal frequency responses of the circuits. The phase noise is also a barrier to the implementation of full duplex communication (Khojastepour et al. 2015). These impose a strong and significant performance limitation to the self-interference cancellation (Syrjala et al. 2014). The phase noise limits the combined digital and analog self interference cancellation. When the residual self-interference is uncorrelated with the self interference signal, the phase noise would be a dominating factor and this would prevent the concentrated digital canceller. Along with these issues, the full duplex MAC layer protocol needs further research, before it could be applied to full-duplex communication (Ali et al. 2014). Some of the main challenges faced by the wireless networks is the presence of the hidden terminals, also referred to as the hidden terminal problem. Moreover, loss of throughput as a result of congestion and end-to-end delays has to be mitigated in order to implement the full duplex communication successfully. Moreover, researches have highlighted that the full duplex communication might not always outperform the half duplex counterpart.   

Performance constrained by self interference: In case of full duplex devices, the input signal of RA, is of lower power, than the magnitude of the self interference signal that was imposed by the TA output of the device (Sugiyama et al. 2014). Hence, the self interference that was imposed on the RA, by the TA would be drowned out by the weak input signal, and thus the full duplex gain was degraded.

Adaptive Signal inversion cancellation: The design of this full duplex communication system is based on simple observation. Any of the full duplex devices that invert a signal through adjustment of the phases is likely to encounter a constraint in the bandwidth. This leads to maximum cancellation. In order to cancel beyond this bound, a full duplex device needs to obtain an inverse signal. This inverse signal is then combined with the transmitted signal such that the self-interference is completely cancelled. A radio needs to invert a signal without any adjustment in its phase (Lawry et al. 2016). A balanced-to-unbalanced converter is used to obtain the inverse of the self-interference signal and this inverted signal is used to cancel the interference. Thus, in order to cancel the self-interference signal, the radon needs to combine the negative signals to match with the self-interference (Ali et al. 2014). This design needs to adapt to the changes for the wireless environment. The basic approach needed for this is to estimate the delay as well as the attenuation of the self-interference signal. The adaptation algorithm would adjust the delay as well as the attenuation, such that the residual energy is minimized (Korpi et al. 2014).

The full duplex communication is simple to accomplish in theory. However, in order to impart full duplex communication in personal communication, various challenges are faced (Nguyen et al. 2014).

Degradation of the reliability of the link: the full duplex communication mode suffers from the degradation of the reliability of the link regardless of the SNR. Without invoking the digital interference, a low reliability of the link, as low as 67% might be attained by the full duplex mode of communication (Korpi et al. 2014).

Suffering due to higher PLR : When compared with the half duplex mode of communication, the full duplex mode needs to process double the number of data packets because of its concurrent transmission (Bharadia and Katti 2014). In this way full duplex mode prompting to a higher PLR than the half duplex mode is suffered.

Higher buffer size needed:  In order to reduce or mitigatethe PLR of the full duplex mode, an adequately large buffer is needed for enabling the data packets to be sent to the receiver. If sufficiently large buffer is not provided, then some of the packets would be discarded as a result of queue overflow (Debaillie et al. 2014). Since the impacts of the loss of the packets are more serious in the full duplex mode, hence a bigger buffer support is required than the one needed in the half duplex mode.  

In the integrated FD transceiver, the scaling of cancellation of the bandwidth up to a level of 80MHz and beyond that level for the emerging wireless standards have been observed. Handling of powerful self-interference at the receiver’s input (Di et al. 2014).

CSMA is founded on the distributed 802.11 protocol. Hence, it is essential to understand the performance of these protocols, as the network is shared between the users of the full duplex communication and the half duplex communication.      

In addition to the above mentioned implementation challenges, the nonlinear distortion also need to be considered. The self interference signal being stronger than the original signal of interest, the component at the receiver end needs to be linear to avoid distortion of the transmitted signal (Aryafar and Keshavarz-Haddad 2015). The non-linearity at the transmitter chain is a major issue in terms of self interference cancellation. Moreover, the analog to digital converter has a limited dynamic range, and with the strong self interference signal coming in, would occupy most of the bandwidth (Kang and Sun 2015). This would result in decrease of the resolution of the signal of interest. This, along with high transmission power, is intensified, and the overall performance of the full-duplex trans-receiver is significantly affected (Tabassum et al. 2015). The full duplex trans-receiver model is a direct conversion model, with full duplex operation. The attenuation of the self interference signal is done at two different stages. These are: the cancellation of RF mitigates the self interference before it enters the receiver chain (Tak-Ki et al. 2014). The digital cancellation suppresses the digital domain.

In order to find the attenuation setting in the real time systems, in order to optimize cancellation, the following algorithm is to be implemented step by step. The stages are as follows:

1. The frequency response of the self interference is to be ca;culated using the WiFi. This process is simple since there are two OFDM symbols. This is a part of the baseband decoding. FFT measure could be performed to measure the frequency response.
2. The integer linear optimization problem is to be solved. After this, the random rounding process is used to find a solution to the attenuator setting. This achieves the required cancellation of 60dB (Ali et al. 2014). This algorithm intent to reduce the search space of the attenuator value of a polynomial set. This is compared to the exponential search space. This is due to the fact that it provides the required cancellation of the self interference. This process cancels the required 60dB (Han et al. 2016). The intuition behind this algorithm is to reduce the search space of the attenuator value. This algorithm is mainly used to look up for the frequency responses of the full duplex mode. In order to improve the algorithm to further cancel the self interference, additional gradient descent stages might be applied to this algorithm. The prototype of the full duplex radio has been given below.

Figure 6: Experimental set-up of full-duplex transceiver

(Source: Heino et al. 2015)

The focus of this  algorithm is on two metrics in order for the evaluation of the cancellation of the self-interference. These are increase in the noise floor and loss of SNR. In this, the presence of the residual interference after the cancellation of the self-interference, an increase in the level of the noise floor from the signal that is received is observed (Heino et al. 2015). The SNR loss refers to the decrease in the signal to noise ratio that is received when a radio operates in the full duplex mode. The SNR of the received signal is of the radio, when it operates in the half duplex mode, is calculated (Cheng et al. 2016). The SNR that is obtained from the received signal, as the radio operates in the full duplex, mode is also calculated. Finally, the differences between these SNR values are considered as the loss in the SNR. In order to achieve balun cancellation, a balun transformer is used to invert a duplicate of the transmitted signal (Ju et al. 2015). The attenuation and the delay is adjusted using attenuators that are programmable. Moreover, this design also uses two antennas that are separated from each other by 20cm. this design is implemented and optimized to produce the best performance (Hong et al. 2016). The rice design uses an extra transmit chain along with the primary transmit chain. This extra transmit chain generates a cancellation signal that could be combined with the signal that is obtained from the receiving chain. This is useful in cancellation of the self-interference. This design is incorporated by the use of by using an extra signal generator as the extra transmit chain for the purpose of cancellation.

However, none of the designs of the full duplex communication mode is successful in cancellation of the self-interferences. For each of the power and locations, 20 runs are conducted and an average of these runs is done. The SNR loss in the received signal in the full duplex mode provides evidence to the amount of cancellation and the increase in the noise floor. In order to prove this, two nodes that are capable of operation in a full duplex mode is set up (Sarret et al. 2015). The first node sends 20 WiFi packets to each other. This is done in a half duplex mode. After this 20 more packets are sending across in full duplex mode. Finally, after 20 packets each is send in the half duplex and full duplex modes, the loss in SNR is calculated. After the experiment is conducted, the loss in SNR is plot as a function of the half duplex SNR. Thus, it has been observed that interference cancellation is not affected by the received signals (Kim et al. 2015). Thus, the SNR loss is less than 1dB. This highlights the fact that in case offull duplex mode, the received signal needs to retain the identical throughput as found in the half duplex mode.

The constellation and the bandwidth have a strong impact on the full duplex communication. Two nodes that are able to communicate in the full duplex mode are set up and the constellation is varied. After this, the SNR loss is calculated. After this the constellation is fixed to 64-QAM and the band width is varied from 20 to 40 MHz. The cancellation techniques have no assumptions about the bandwidth, constellation of any other parameter. Thus, the constellation and bandwidth makes no assumptions.                           

The most important factor in the analog cancellation circuit board is the count of delay lines that are fixed. Since the circuit boards are not flexible, hence the number of lines in increments of one (Cui et al. 2014). The plot needs to be read as the power of self –interference after the phase of analog cancellation as a function of the frequency. The difference of the higher capacity is of 16 lines in the cancellation signal along with cancellation of the self reflection of the main self-interference components that leaks through the circulator (Duarte et al. 2014). As the full duplex node transmits, the responses from the antenna and the circulator components have two leak chain components. Since both these components travel in different paths, hence they undergo different delays from the measurements of the time domain. The first delay was found to be 400 picoseconds. The reflected component was found to be1.4 nanoseconds.       

As the 62dB of the cancellation of the analog component takes place, the digital components need to be reduced by 16dB from the non-linear self interference and 48dB of the linear component. For conduction of this experiment, the analog cancellation circuit is tuned to provide a cancellation of 62dB. More components re then added to the digital cancellation design (Yilmaz and Dehollain 2014). First the linear digital cancellation that cancels the linear component of the environment is considered. After this, when the capability of the model is added to the non-linear model, the non-linear cancellation is obtained. Thus, the cancellation of the two variants is obtained by the technique of digital cancellation.

With the changes in the environment, the levels of the drop of the cancellation values of the attenuators that is used, would be off with respect to the new conditions. The digital cancellation is tuned using special tuning period (Sahai et al. 2013). During this period no data is transmitted. This experiment is conducted using indoor environment with various WiFi radios that were moving around. However, it is to be noted that an indoor environment is not suitable for full duplex communication experiments. This is due to the reason that there are a huge number of reflectors in the indoor environment. The full duplex node is placed and the analog cancellation is conducted (Li et al. 2014). The WiFi preamble is used to determine the initial settings of the attenuator that matches the frequency response of the circulator as well as the antenna. This experiment is conducted several times in different environments. The difference of the node placement is also changed in each time the experiment is conducted.  

The various approaches in order to reduce self-interference for the IBFD terminals includes the propagation domain, analog-circuit domain as well as the digital domain. In order to understand the approaches for reduction of self-interference, it is essential to analyze the design of the prototypical IBFD terminal (Ali et al. 2014). The design of a separate antenna IBFD terminal has been presented here. A shared antenna IBFD terminal is similar, except that each of the transmit- receive antenna pair, is replaced with circulator attached. The IBFD terminal accepts a transmit bit stream that is coded a well as modulated in the digital domain, thus producing a separate digital baseband signal (Everett et al. 2015). Each of the digital signals is converted with digital to analog converter. This helps in converting to a higher carrier frequency. However, in the practical world, this would produce various non-idealities in the transit signal. This leads to a small difference between the intended signal and actual signal. Over the same band of frequency, the IBFD terminal functions as a receiver. The signal that is received by each antenna is then put through a separate hardware, with includes a low-noise amplifier, along with analog to digital converter and down converter. The received signal could be decomposed into three separate components. These are the desired signal, the self interference propagation directly from the transit chain and the self interference reflection (Zhou et al. 2014). While the path direct self interference could be accurately characterized offline, the system design along with the reflected path cannot, as they are dependent on the environmental effects that are unpredictable.             

Conclusion      

It has been observed that with the rise in the use of mobiles among the people around the world, the need for full duplex communication modes became essential. Full duplex communication refers to the two way communication that allows each of the devices to send data packets along with receiving acknowledgements simultaneously. However, though theoretically it might seem to be simple, yet it has a number of implementation challenges. The implementation challenges include self-interference, degradation of the reliability of the link, suffering due to higher PLR as well as need for higher buffer size. The various challenges have been highlighted in this assignment. Along with the various challenges, the way these challenges could be reduced is also mentioned in details. Thus, the efficiency of the full-duplex communication makes it inevitable to be used, even with a number of challenges.

References

Ahmed, E., Eltawil, A.M. and Sabharwal, A., 2013, November. Self-interference cancellation with nonlinear distortion suppression for full-duplex systems. In Signals, Systems and Computers, 2013 Asilomar Conference on (pp. 1199-1203). IEEE.

Ali, S., Rajatheva, N. and Latva-aho, M., 2014, June. Full duplex device-to-device communication in cellular networks. In Networks and Communications (EuCNC), 2014 European Conference on (pp. 1-5). IEEE.

Aryafar, E. and Keshavarz-Haddad, A., 2015, April. FD 2: A directional full duplex communication system for indoor wireless networks. In Computer Communications (INFOCOM), 2015 IEEE Conference on (pp. 1993-2001). IEEE.

Bharadia, D., McMilin, E. and Katti, S., 2013. Full duplex radios. ACM SIGCOMM Computer Communication Review, 43(4), pp.375-386.

Bharadia, D. and Katti, S., 2014. Full duplex MIMO radios. Self, 1(A2), p.A3.

Cheng, H., Wang, R. and Du, Y., Huawei Technologies Co., Ltd., 2016. Method and full-duplex communication device for acquiring channel response of self-interfering channel. U.S. Patent 9,504,048.

Cheng, W., Zhang, X. and Zhang, H., 2015. Full-duplex spectrum-sensing and MAC-protocol for multichannel nontime-slotted cognitive radio networks. IEEE Journal on Selected Areas in Communications, 33(5), pp.820-831.

Chin, W.H., Fan, Z. and Haines, R., 2014. Emerging technologies and research challenges for 5G wireless networks. IEEE Wireless Communications, 21(2), pp.106-112.

Chung, M., Sim, M.S., Kim, J., Kim, D.K. and Chae, C.B., 2015. Prototyping real-time full duplex radios. IEEE Communications Magazine, 53(9), pp.56-63.

Cui, H., Ma, M., Song, L. and Jiao, B., 2014. Relay selection for two-way full duplex relay networks with amplify-and-forward protocol. IEEE transactions on wireless communications, 13(7), pp.3768-3777.

Debaillie, B., van den Broek, D.J., Lavin, C., van Liempd, B., Klumperink, E.A., Palacios, C., Craninckx, J., Nauta, B. and Pärssinen, A., 2014. Analog/RF solutions enabling compact full-duplex radios. IEEE Journal on Selected Areas in Communications, 32(9), pp.1662-1673.

Di, B., Bayat, S., Song, L. and Li, Y., 2014, April. Radio resource allocation for full-duplex OFDMA networks using matching theory. In Computer Communications Workshops (INFOCOM WKSHPS), 2014 IEEE Conference on (pp. 197-198). IEEE.

Duarte, M., Sabharwal, A., Aggarwal, V., Jana, R., Ramakrishnan, K.K., Rice, C.W. and Shankaranarayanan, N.K., 2014. Design and characterization of a full-duplex multiantenna system for WiFi networks. IEEE Transactions on Vehicular Technology, 63(3), pp.1160-1177.

ElSawy, H., AlAmmouri, A., Amin, O. and Alouini, M.S., 2016, May. Can uplink transmissions survive in full-duplex cellular environments?. In European Wireless 2016; 22th European Wireless Conference; Proceedings of (pp. 1-6). VDE.

Everett, E., Sahai, A. and Sabharwal, A., 2014. Passive self-interference suppression for full-duplex infrastructure nodes. IEEE Transactions on Wireless Communications, 13(2), pp.680-694.

Gallinaro, G., Tirrò, E., Cecca, F.D., Migliorelli, M., Gatti, N. and Cioni, S., 2014. Next generation interactive S?band mobile systems: challenges and solutions. International Journal of Satellite Communications and Networking, 32(4), pp.247-262.

Goyal, S., Liu, P., Panwar, S.S., Difazio, R.A., Yang, R. and Bala, E., 2015. Full duplex cellular systems: will doubling interference prevent doubling capacity?. IEEE Communications Magazine, 53(5), pp.121-127.

Han, X., Huo, B., Shao, Y. and Zhao, M., 2016, September. Optical RF self-interference cancellation for full-duplex communication using phase modulation and sideband filtering. In Optical Communications and Networks (ICOCN), 2016 15th International Conference on (pp. 1-3). IEEE.

Heino, M., Korpi, D., Huusari, T., Antonio-Rodriguez, E., Venkatasubramanian, S., Riihonen, T., Anttila, L., Icheln, C., Haneda, K., Wichman, R. and Valkama, M., 2015. Recent advances in antenna design and interference cancellation algorithms for in-band full duplex relays. IEEE Communications Magazine, 53(5), pp.91-101.

Hong, S., Katti, S. and Mehlman, J., The Board Of Trustees Of The Leland Stanford Junior University, 2016. Systems and methods for full-duplex signal shaping. U.S. Patent Application 15/133,175.

Hossain, E. and Hasan, M., 2015. 5G cellular: key enabling technologies and research challenges. IEEE Instrumentation & Measurement Magazine, 18(3), pp.11-21.

Hui, J.W., Vasseur, J.P. and Hong, W., Cisco Technology, Inc., 2016. Full-duplex capacity allocation for OFDM-based communication. U.S. Patent 9,240,913.

Ju, H. and Zhang, R., 2014. Optimal resource allocation in full-duplex wireless-powered communication network. IEEE Transactions on Communications, 62(10), pp.3528-3540.

Ju, H., Chang, K. and Lee, M.S., 2015, July. In-band full-duplex wireless powered communication networks. In Advanced Communication Technology (ICACT), 2015 17th International Conference on (pp. 23-27). IEEE.

Kang, X., Ho, C.K. and Sun, S., 2015. Full-duplex wireless-powered communication network with energy causality. IEEE Transactions on Wireless Communications, 14(10), pp.5539-5551.

Khojastepour, M.A., Barghi, S., Sundaresan, K. and Rangarajan, S., Nec Laboratories America, Inc., 2015. Multiple-input multiple-output wireless communications with full duplex radios.

Kim, S. and Stark, W., 2014, February. Full duplex device to device communication in cellular networks. In Computing, Networking and Communications (ICNC), 2014 International Conference on (pp. 721-725). IEEE.

Kim, W., Song, T., Kim, T. and Pack, S., 2015. Medium access control protocols for full-duplex communications in WLAN systems: approaches and challenges. The Journal of Korean Institute of Communications and Information Sciences, 40(7), pp.1276-1285.

Kim, D., Lee, H. and Hong, D., 2015. A survey of in-band full-duplex transmission: From the perspective of PHY and MAC layers. IEEE Communications Surveys & Tutorials, 17(4), pp.2017-2046.

Korpi, D., Riihonen, T., Syrjälä, V., Anttila, L., Valkama, M. and Wichman, R., 2014. Full-duplex transceiver system calculations: Analysis of ADC and linearity challenges. IEEE Transactions on Wireless Communications, 13(7), pp.3821-3836

Korpi, D., Tamminen, J., Turunen, M., Huusari, T., Choi, Y.S., Anttila, L., Talwar, S. and Valkama, M., 2016. Full-duplex mobile device: pushing the limits. IEEE Communications Magazine, 54(9), pp.80-87.

Korpi, D., Riihonen, T. and Valkama, M., 2015, March. Achievable rate regions and self-interference channel estimation in hybrid full-duplex/half-duplex radio links. In Information Sciences and Systems (CISS), 2015 49th Annual Conference on (pp. 1-6). IEEE.

Korpi, D., Riihonen, T., Syrjälä, V., Anttila, L., Valkama, M. and Wichman, R., 2014. Full-duplex transceiver system calculations: Analysis of ADC and linearity challenges. IEEE Transactions on Wireless Communications, 13(7), pp.3821-3836.

Lawry, T.J., Saulnier, G.J., ASHDOWN, J.D., Scarton, H.A. and Gavens, A., Rensselaer Polytechnic Institute, 2016. Full-duplex ultrasonic through-wall communication and power delivery system with frequency tracking. U.S. Patent 9,455,791.

Li, W., Lilleberg, J. and Rikkinen, K., 2014. On rate region analysis of half-and full-duplex OFDM communication links. IEEE Journal on Selected Areas in Communications, 32(9), pp.1688-1698.

Liao, Y., Song, L., Han, Z. and Li, Y., 2015. Full duplex cognitive radio: a new design paradigm for enhancing spectrum usage. IEEE Communications Magazine, 53(5), pp.138-145.

Liu, G., Yu, F.R., Ji, H., Leung, V.C. and Li, X., 2015. In-band full-duplex relaying: A survey, research issues and challenges. IEEE Communications Surveys & Tutorials, 17(2), pp.500-524.

Liu, G., Yu, F.R., Ji, H., Leung, V.C. and Li, X., 2015. In-band full-duplex relaying: A survey, research issues and challenges. IEEE Communications Surveys & Tutorials, 17(2), pp.500-524.Mahmood, N.H., Berardinelli, G., Tavares, F.M. and Mogensen, P., 2015, May. On the potential of full duplex communication in 5G small cell networks. In Vehicular Technology Conference (VTC Spring), 2015 IEEE 81st (pp. 1-5). IEEE.

Mahmood, N.H., Berardinelli, G., Mogensen, P. and Frederiksen, F., 2015. Throughput analysis of full duplex communication with asymmetric traffic in small cell systems. ICWMC 2015, p.68. 

Nguyen, D., Tran, L.N., Pirinen, P. and Latva-aho, M., 2014. On the spectral efficiency of full-duplex small cell wireless systems. IEEE Transactions on wireless communications, 13(9), pp.4896-4910.

Prendergast, A.J., Humphrey, J.H., Mutasa, K., Majo, F.D., Rukobo, S., Govha, M., Mbuya, M.N., Moulton, L.H. and Stoltzfus, R.J., 2015. Assessment of environmental enteric dysfunction in the SHINE trial: methods and challenges. Clinical Infectious Diseases, 61(suppl 7), pp.S726-S732.

Rodriguez, L.J., Tran, N.H., Duong, T.Q., Le-Ngoc, T., Elkashlan, M. and Shetty, S., 2015. Physical layer security in wireless cooperative relay networks: State of the art and beyond. IEEE Communications Magazine, 53(12), pp.32-39.

Sabharwal, A., Schniter, P., Guo, D., Bliss, D.W., Rangarajan, S. and Wichman, R., 2014. In-band full-duplex wireless: Challenges and opportunities. IEEE Journal on Selected Areas in Communications, 32(9), pp.1637-1652.

Sabharwal, A., Schniter, P., Guo, D., Bliss, D.W., Rangarajan, S. and Wichman, R., 2014. In-band full-duplex wireless: Challenges and opportunities. IEEE Journal on Selected Areas in Communications, 32(9), pp.1637-1652.

Sahai, A., Patel, G., Dick, C. and Sabharwal, A., 2013. On the impact of phase noise on active cancelation in wireless full-duplex. IEEE transactions on Vehicular Technology, 62(9), pp.4494-4510.

Sarret, M.G., Catania, D., Berardinelli, G., Mahmood, N.H. and Mogensen, P., 2015, September. Full duplex communication under traffic constraints for 5G small cells. In Vehicular Technology Conference (VTC Fall), 2015 IEEE 82nd (pp. 1-5). IEEE.

Song, L., Li, Y. and Han, Z., 2015. Resource allocation in full-duplex communications for future wireless networks. IEEE Wireless Communications, 22(4), pp.88-96.

Sugiyama, Y., Tamaki, K., Saruwatari, S. and Watanabe, T., 2014, January. A wireless full-duplex and multi-hop network with collision avoidance using directional antennas. In Mobile Computing and Ubiquitous Networking (ICMU), 2014 Seventh International Conference on (pp. 38-43). IEEE.

Syrjala, V., Valkama, M., Anttila, L., Riihonen, T. and Korpi, D., 2014. Analysis of oscillator phase-noise effects on self-interference cancellation in full-duplex OFDM radio transceivers. IEEE Transactions on Wireless Communications, 13(6), pp.2977-2990.

Tabassum, H., Hossain, E., Ogundipe, A. and Kim, D.I., 2015. Wireless-powered cellular networks: Key challenges and solution techniques. IEEE Communications Magazine, 53(6), pp.63-71.

Tak-Ki, Y.U., Chang, Y.B., Kim, Y.S., Son, J.J., Han, S.H., Yun, S.B., Lee, J.H., Hwang, S.S. and Jung, S.Y., Samsung Electronics Co., Ltd, 2014. Network entry apparatus and method for relay station using full duplex in mobile communication system. U.S. Patent 8,837,390.

Tang, A. and Wang, X., 2015. A-duplex: Medium access control for efficient coexistence between full-duplex and half-duplex communications. IEEE Transactions on Wireless Communications, 14(10), pp.5871-5885.

Thilina, K.M., Tabassum, H., Hossain, E. and Kim, D.I., 2015. Medium access control design for full duplex wireless systems: challenges and approaches. IEEE Communications Magazine, 53(5), pp.112-120.

Vermeulen, T. and Pollin, S., 2014, December. Energy-delay analysis of full duplex wireless communication for sensor networks. In Global Communications Conference (GLOBECOM), 2014 IEEE (pp. 455-460). IEEE.

Wang, L., Tian, F., Svensson, T., Feng, D., Song, M. and Li, S., 2015. Exploiting full duplex for device-to-device communications in heterogeneous networks. IEEE Communications Magazine, 53(5), pp.146-152.

Yilmaz, G. and Dehollain, C., 2014. Single frequency wireless power transfer and full-duplex communication system for intracranial epilepsy monitoring. Microelectronics Journal, 45(12), pp.1595-1602.

Yun, J.H., 2016. Intra and inter-cell resource management in full-duplex heterogeneous cellular   networks. IEEE Transactions on Mobile Computing, 15(2), pp.392-405.

Zhang, Z., Chai, X., Long, K., Vasilakos, A.V. and Hanzo, L., 2015. Full duplex techniques for 5G networks: self-interference cancellation, protocol design, and relay selection. IEEE Communications Magazine, 53(5), pp.128-137.

Zhang, Z., Long, K., Vasilakos, A.V. and Hanzo, L., 2016. Full-duplex wireless communications: challenges, solutions, and future research directions. Proceedings of the IEEE, 104(7), pp.1369-1409.

Zhou, M., Cui, H., Song, L. and Jiao, B., 2014. Transmit-receive antenna pair selection in full duplex systems. IEEE Wireless Communications Letters, 3(1), pp.34-37.

Zlatanov, N., Ikhlef, A., Islam, T. and Schober, R., 2014. Buffer-aided cooperative communications: opportunities and challenges. IEEE Communications Magazine, 52(4), pp.146-153.