Contention-based Random Access For Massive Connections In 5G

The fifth generation(5G)wireless network faces many unprecedented challenges.Among them,wireless channel congestion due to the soaring number of concurrent access requests is of prominent importance.Non-orthogonal multiple access schemes have been widely studied in industrial researches for its potential in coping with massive connections.Meanwhile,massive MIMO,which is considered as a key enabling technology in 5G,has very high spatial resolution and can expand system capacity significantly thanks to its large number of antennas.In this paper,random access schemes for massive connections in 5 G is discussed and two different aspects,including non-orthogonal coded access and random access in massive MIMO system,are analyzed.Non-orthogonal coded access(NOCA)is a candidate for 5G radio access technology.In the scheme,each data symbol is spread by non-orthogonal sequence and mapped to multiple subcarriers of OFDMA.Every User chooses its spreading sequence from a non-orthogonal codebook and spreads its data onto the time-frequency resource shared by all users.Spreading sequences serve to identify data from different users.Non-orthogonality between sequences augments the number of sequences available and effectively reduces the probability of user collision.At the same time,however,the non-orthogonal codebook inevitably introduces interference between multiple users.To resolve the latter issue,an advanced receiver is implemented to perform user detection and interference cancellation(IC)iteratively.Then,two types of simulations are designed to evaluate the capability of NOCA under heavy loaded scenarios.Link level simulations which focus on radio link performance and system level simulations which observe network performance are carried out.NOCA schemes display remarkable adaptability to heavy load and robustness against the interference in both types of simulations.After that,taking into account the nature of iterative IC receivers,an optimized power control(PC)scheme is proposed to further improve the performance of NOCA.The proposed uplink power allocation scheme is based on partial compensation of large-scale fading of users.By creating received power variations,it not only enhances the system capacity and throughput but also leads to lower power consumptions.Computer simulations prove that NOCA is capable of supporting 4 times more connections than OFDMA in contention-based transmission at 1%packet error rate.Random access scheme in massive MIMO is then studied in support of more connected devices.Making use of the channel hardening characteristic of massive MIMO system,it allows multiple users to share the same pilot sequence in uplink transmission.Base station estimates the composite channel of all users choosing the pilot,and restore transmitted data utilizing the channel estimation.In the original received signal,data from different users go through vectorial complex Gaussian channels.After maximum ratio combining,the vectorial channels are transformed into real positive scalar channels.Therefore,user data can be successively recovered by iterative IC,similar to the general idea of power-domain multiplexing.Decoding of user data relies on received power gaps between users,the random access problem is thus converted to an uplink power allocation problem.In this paper,theoretical optimal power control schemes are derived based on outage probability and decoding error rate,under the assumption that global information is ideally known by all participants.Then based on large-scale fading partial compensation and probabilistic multi-level selection,two power allocation schemes are put forward so as to enable users to make independent PC decisions while achieving near-optimal performance.It is proved through simulations that theoretical power allocation provides the best performance among schemes.The large-scale fading partial compensation based scheme,by contrast,ensures reasonable performance with several advantages including distributed implementation and reduced power requirement,which could be a good start for following researches in the domain.