Access the full text.
Sign up today, get DeepDyve free for 14 days.
A Abdelnasser, E Hossain, DI Kim (2014)
Clustering and resource allocation for dense femtocells in a two-tier cellular OFDMA networkIEEE Trans. Wireless Commun., 13
S Sarkar, S Chatterjee, S Misra (2015)
Assessment of the suitability of fog computing in the context of internet of thingsIEEE Trans. Cloud Comput., 6
H Aissi, C Bazgan, D Vanderpooten (2005)
Complexity of the min-max and min-max regret assignment problemsOper. Res. Lett., 33
A Mehrotra, MA Trick (2010)
A branch-and-price approach for graph multi-coloringOper. Res. Comput. Sci. Interfaces, 37
Z Chen, Y Tang (2017)
Computer Science and Education (ICCSE), 2017 12th International Conference On
S Bassoy, H Farooq, MA Imran, A Imran (2017)
Coordinated multi-point clustering schemes: a surveyIEEE Commun. Surv. Tutorials, 19
B Han, W Wang, Y Li, M Peng (2013)
Investigation of interference margin for the co-existence of macrocell and femtocell in orthogonal frequency division multiple access systemsIEEE Syst. J., 7
Z Sun, R Liu, W Wang (2013)
Joint time-frequency domain cyclostationarity-based approach to blind estimation of OFDM transmission parametersEURASIP J. Wireless Commun. Netw., 2013
S Chen, F Qin, B Hu, X Li, Z Chen (2016)
User-centric ultra-dense networks for 5G: challenges, methodologies, and directionsIEEE Wireless Commun., 23
C Yang, J Li, M Guizani (2016)
Cooperation for spectral and energy efficiency in ultra-dense small cell networksIEEE Wireless Commun., 23
S Zou, F Yang, Y Tang, L Xiao (2015)
The resource mapping algorithm of wireless virtualized networks for saving energy in ultradense small cellsMob. Inf. Syst., 2015
JG Andrews, S Buzzi, W Choi, SV Hanly, A Lozano, AC Soong, JC Zhang (2014)
What will 5G be?IEEE J. Selected Areas Commun., 32
S Lin, H Tian (2013)
Wireless Communications and Networking Conference (WCNC), 2013 IEEE
A Ghosh, R Ratasuk, B Mondal, N Mangalvedhe, T Thomas (2010)
LTE-advanced: next-generation wireless broadband technology [invited paper]IEEE Wireless Commun., 17
Y Dong, Z Chen, P Fan, KB Letaief (2016)
Mobility-aware uplink interference model for 5G heterogeneous networksIEEE Trans. Wireless Commun., 15
A Hatoum, R Langar, N Aitsaadi, R Boutaba, G Pujolle (2014)
Cluster-based resource management in OFDMA femtocell networks with QoS guaranteesIEEE Trans. Veh. Technol., 63
O Semiari, W Saad, S Valentin, M Bennis, B Maham (2014)
Acoustics, Speech and Signal Processing (ICASSP), 2014 IEEE International Conference On
Y Li, Z Feng, S Chen, Y Chen, D Xu, P Zhang, Q Zhang (2011)
Radio resource management for public femtocell networksEURASIP J. Wireless Commun. Netw., 2011
T Lotfollahzadeh, S Kabiri, H Kalbkhani, MG Shayesteh (2016)
Femtocell base station clustering and logistic smooth transition autoregressive-based predicted signal-to-interference-plus-noise ratio for performance improvement of two-tier macro/femtocell networksIET Signal Process., 10
Y Zhang, S Wang, J Guo (2015)
Communications in China (ICCC), 2015 IEEE/CIC International Conference On
M Bennis, SM Perlaza, P Blasco, Z Han, HV Poor (2013)
Self-organization in small cell networks: A reinforcement learning approachIEEE Trans. Wireless Commun., 12
I Chih-Lin, J Huang, R Duan, C Cui, JX Jiang, L Li (2014)
Recent progress on C-RAN centralization and cloudificationIEEE Access, 2
J Zheng, Y Cai, A Anpalagan (2015)
A stochastic game-theoretic approach for interference mitigation in small cell networksIEEE Commun. Lett., 19
Y Li, Z Feng, D Xu, Q Zhang, H Tian (2011)
Optimisation approach for femtocell networks using coordinated multipoint transmission techniqueElectron. Lett., 47
TA Weiss, FK Jondral (2004)
Spectrum pooling: an innovative strategy for the enhancement of spectrum efficiencyIEEE Commun. Mag., 42
A Damnjanovic, J Montojo, Y Wei, T Ji, T Luo, M Vajapeyam, T Yoo, O Song, D Malladi (2011)
A survey on 3GPP heterogeneous networksIEEE Wireless Commun., 18
Y Wang, P Zhang (2005)
Radio resource management
In a 5G ultra-dense network, dynamic network topology and traffic patterns lead to excessive system overhead and complex radio resource conflicts. The cloud radio access network and the fog computing have the advantages of high computation capabilities and low transmission delays. Therefore, by taking full advantage of these two characteristics, this study proposes a novel radio resource coordination and scheduling scheme in an ultra-dense cloud-based small cell network. Interference among small cells (or remote radio heads) can be avoided by implementing centralized cooperative processing in the base band unit pool in advance. Resource sharing in coordination and transfer depend on fog computing to relieve the overloaded cloud processing platform and reduce transmission delays, thereby maximizing resource utilization and minimizing system overhead when the network topology and number of users change dynamically. The simulation shows that the proposed scheme can increase the system throughput by 20% compared with the clustering-based algorithm; it can also increase system throughput by 33% compared with the graph coloring algorithm, decrease the signaling overhead by about 50%, and improve network’s quality of service. Keywords: Ultra-dense network (UDN), The cloud radio access network (Cloud-RAN), Fog computing, Resource coordination, 5th generation (5G), Small cell 1 Introduction interference problem includes radio resource conflict and Given current demand, broadband mobile data is cell interference, whereas the system overhead problem expected to be ubiquitously available. The industry has affects delay, signaling interaction, and computational predicted a 1000-fold increase in mobile data traffic abilities. within the next decade. Ultra-dense network (UDN) The ultra-dense cloud-based small cell network com- deployment appears promising for improving network bines cloud computing and fog computing in the devel- capacity [1, 2]. Traditional operators must deploy more opment of a large number of small cells. The cloud infrastructure to address myriad challenges associated radio access network (Cloud-RAN) presents a promising with data proliferation, which increases total costs signif- method for alleviating both capital and interference in icantly [3, 4]. Radio resource management and schedul- UDNs, while providing high energy efficiency and capac- ing in UDNs comprise an important means of network ity. Virtualized network technology is a key technology capacity promotion; however, because a larger number of resource management and scheduling in the fifth gen- of small cells are needed to promote higher data capac- eration (5G) mobile communication networks [5, 6]. Vir- ity, problems related to interference and system overhead tualized network technology allows why Cloud-RAN to in the current UDN are much more severe than those integrate centralization and virtualization into its archi- in existing cellular mobile communication networks. The tecture. The resources can be better managed and dynam- ically coordinated on demand on a pool level because *Correspondence: tyl@xmu.edu.cn Cloud-RAN centralizes all base band units (BBU) to form Equal contributors a pool, and remote radio heads (RRH) provide basic wire- Department of Communication Engineering, Xiamen University, Xiamen 361005, China less signal coverage. In practice, the front haul and back Full list of author information is available at the end of the article haul of Cloud-RAN are often constrained by capacity and © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 2 of 15 delays. Fog computing appears to be charged with most algorithm to be less complex. According to the clustered computation assignments and services from the cloud small cells and predicted signal to interference plus noise processing platform in the network’s edge, which eases ratio(SINR), theauthors of [18] proposed a power control overload [7]. Clearly, Cloud-RAN with the better compu- scheme applied to the downlink in the small cell network tational capability (10 times larger than that of the fog and found this method to lower the likelihood of an out- RAN) and the fog computing with a lower transmission age while improving throughput significantly compared delay are complementary in UDNs. to previous methods. However, these methods require The main issue faced by resource management and considerable signaling interaction between small cells to scheduling is the reduction of inter-cell interference in 5G complete the resource allocation. UDNs. Interference includes inter-small cell interference To adapt to the various demands of radio traffic, net- (ISI) and macro-small cell interference (MSI) [8]. MSI can work virtualization and UDNs represent key technologies be solved by using different frequencies; for instance, the for future 5G wireless networks. In the virtualization macro cell uses a low frequency, and the small cell uses architecture of 5G wireless networks, significant changes a high frequency [9]. The third generation partnership occur in the signaling interaction of virtual cells and the project (3GPP) R8 and R9 minimizes inter-cell interfer- management mode of multi-dimension wireless networks. ence in the macro and maximizes spectrum efficiency Based on interference coordination in traditional cells, through frequency reuse planning. Due to the unplanned the current network needs a more flexible and efficient deployment of small cells, operators have little control sharing of radio resources to realize their coordinated over small cells’ position and thus cannot devise frequency allocation, including in the time, frequency, spatial, and plans in advance. Therefore, the reasonable allocation power domains [19, 20]. The wireless network virtualiza- of radio resources in small cells, such as frequency, has tion technology forms a virtual network in isolation for become the focus of current research to reduce the inter- different users by sharing the network’s physical infras- ference. Coordinated multi-point (CoMP) is a key method tructure and radio resources to achieve efficient use in of mitigating inter-cell interference that can be applied [5, 6]. A centralized resource pool must be built to achieve to the network architecture, such as control/data plane dynamic sharing and coordination. The resource pool separation architecture and Cloud-RAN [10]. provides the distributed function unit a logically central- Given growing demand for mobile data capacity, espe- ized radio resource and processing resource to decou- cially in urban areas, network densification is inevitable ple resource and network functions, to realize resource [11]. The user equipment in ultra-dense small cells has a sharing and distribution in different network entities, tidal effect; the authors of [12] proposed a mobility-aware and to improve resource utilization overall [21, 22]. In uplink interference model for 5G heterogeneous networks [23, 24], a study conducted on interference coordina- to solve the uplink interference with time variation. The tion in a heterogeneous wireless virtualization network, authors of [13] presented a completely distributed chan- using power control to reduce interference between small nel allocation algorithm based on game theory. Although cells and thereby improving coverage and indoor small these methods improved radio resource utilization, they cells’ capacity. also created larger system overheads. To minimize ISI However, the interference and system overhead remains and ensure users’ quality of service, some academics have more severe in 5G UDNs due to dynamic traffic and used small cell clustering to coordinate cluster mem- the random increase or migration of small cells. Current bers’ resource allocation. Considering graph theory, a resource allocation algorithms for small cells are unable to clustering-based interference coordination heuristic algo- manage poor situations, an urgent problem that requires rithm was proposed in [14–16], which forms clusters an innovative solution. Inspired by the method of collabo- dynamically and divides resource allocation into three ration between small cells in COMP, a resource coordina- phases to minimize ISI: (1) cluster formation; (2) intra- tion approach was proposed in [25]tosolve theproblem. cluster resource allocation and admission control; and (3) Considering Cloud-RAN and fog computing, this study inter-cluster resource contention resolution, in order to presents a novel resource coordination and scheduling minimize ISI. It is worth noting that in this model, the scheme in ultra-dense cloud-based small cells. formationofagivencluster is updatedonlywhenthe The rest of this paper is organized as follows. Section 2 interference caused by the changes in the user number introduces the methods of this paper. Section 3 describes exceeds a given threshold. In [17], authors devised a clus- the system model of Cloud-RAN and fog computing. tering and resource allocation for dense femtocells in a Section 4 details the resource coordination and schedul- two-tier cellular orthogonal frequency division multiplex- ing scheme in ultra-dense small cells. Section 5 describes ing access (OFDMA) network. This solution converts the a resource scheduling example and corresponding analy- interference problem into a mixed integer nonlinear opti- ses of signaling overheads. Scenario setting of simulation mization model, and the simulation results found this are presented in Section 6, and the results and discussion Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 3 of 15 of this paper are proposed in Section 7.Finally,the con- are insufficient, the fog device lacking resources needs to clusion and ending remarks are given in Section 8. either send a request for resources to the system resource pool in the cloud or borrow resources from adjacent fog 2 Methods devices, a process that also aims to minimize the amount Considering Cloud-RAN and fog computing, this study of PRB switching. presents a novel resource coordination and scheduling scheme in ultra-dense cloud-based small cells to improve 3Systemmodel system throughput and reduce system signaling over- Multiple small cells are deployed in the area with a den- head. After network initialization is accomplished, a sity of 177 cells per km , and any two cells whose signals global resource allocation algorithm is used in the cloud interfere with each other are termed “adjacent cell.” This processing platform to allocate resources for each fog study uses the differ-frequency development scenarios of device according to the global signaling request. When macro-small cell 3GPP TR 36.932 proposed; there is no the tidal effect occurred or resources are deficient, the co-channel interference between the macro cell and small fog devices in the edge of the cloud executes the partial cells due to the differ-frequency network. Based on the resource optimization algorithm or resource coordination architecture of Cloud-RAN and fog computing, small cells and scheduling scheme to adjust the number of resources are spliced as BBU and RRH, and the fog device connects between each fog device. Resource coordination between the cloud processing platform via the back haul network fog devices can be divided into three situations: (1) when and connects the corresponding small cells via front haul some small cells power off, the cloud-based system will network. As shown in Fig. 1, a BBU pool is located in a obtain resource of the fog device that these small cells centralized cloud processing platform that is used to man- belong to; (2) when some small cells power on, the cloud age the radio resources in the system resource pool. The adjusts the resources between the fog device, whose small cloud processing platform manages all fog devices, and cells power on, and its adjacent fog device to ensure a min- different fog devices manage small cells in various traffic imal amount of physical resource block (PRB) switching areas. There may be more than two fog devices in sys- in subsequent processes; (3) when fog device resources tem; small cells with the same characteristic (depending Core BBU pool Network Gateway Cloud processing platform Back haul ISI Data Fog device Fog device RRH Front haul Small cell Fig. 1 System network model Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 4 of 15 on traffic and location) connect to the associated fog small cells s are allocated to the UE u , which must be i k device via a front haul network. Different fog devices can greater than or equal to the minimum rate demand R of communicate when resource coordination is required. the UE u . A table of notations for this paper is provided in Table 2. { } Set S = s , ..., s as the set of small cells in the area G; 1 m 4 Resource coordination and scheduling scheme ∗ ∗ m ∈ N , N represents the set of positive integers. B = in small cells {b , ..., b } donates PRBs in the system resource pool, and 1 n The problem of resource allocation in this paper is sim- n ∈ N . U = {u , ..., u } is a set of user equipment (UE) in 1 o ilar to graph coloring problem (GCP). The same PRB a small cell, o ∈ N . Due to the downlink interference, the cannot be assigned to adjacent nodes no matter which rate R UE u obtained based on PRB b in small cell s is i,k,j k j i node the resource allocation starts from, until the PRB resource allocation of all nodes has completed. More- R = W · log (1 + Sinr ),(1) i,k,j i,k,j over, the problems in our paper are different from GCP. Each node has only one color in GCP, but each nodes where W is the bandwidth of PRB, and Sinr represents i,k,j needs to be allocated one or more PRBs in our prob- the SINR UE u received based on PRB b in s , which can k j i lems, so it not only belongs to GCP but also belongs to be calculated by subset selection problem. GCP is one of 33 NP-hard prob- p · g i,k,j i,k,j lems, and subset selection problem has been proven to be Sinr = ,(2) i,k,j NP-hard problem in [26]. Therefore, heuristic algorithm p · g + n l,k,j l,k,j 0 l=1 is needed to solve the problem of resource allocation l=i in this paper. A novel radio resource coordination and scheduling scheme in small cells (including three heuristic where p indicates the transmission power from serving i,k,j algorithms) are proposed to solve the resource alloca- small cell s to UE u based on PRB b , p indicates the i k j l,k,j tion problem. A specific function diagram of the scheme interference power from intruder small cell s to UE u l k is showninFig. 2. Global resource allocation and par- based on PRB b , n stands for the noise power, and g is j 0 i,k,j tial resource allocation are executed in a cloud processing the channel gain, and the channel gain is composed of link platform with increased computational capabilities. The loss and the antenna gain. resource coordination and scheduling is processed in a Set BS = {b s , b s , ...} as the available PRB set of small i 1 i 2 i fog device with larger storage. First, the cloud process- cell s . For any UE u in the network, R is a constant i k k ing platform allocates PRB for each fog device to try to used to indicate the minimum rate demand of UE u .The meet the pre-demands of small cells using minimum PRB constraint that the UE u can access the small cell s at a k i in the system resource pool, thereby reducing the sig- certain position is naling overheads and ensuring interference optimization in the whole network. Then, the cloud processing plat- R R,(3) i,k,j k form updates real-time interference graphs (mentioned b s ∈BS j i i in Section 3.1.2) with topology changes, and a partial where the summation R stands for the cur- resource allocation algorithm is used to adjust the number i,k,j b s ∈BS j i i rent rate that the UE u obtained when all the PRBs in of allocated PRBs in each fog device. A resource request Cloud computing Global resource allocation Partial resource optimization Coordinated resource scheduling Adopt allocated PRB in serving small cell Use remaining PRB in system resource pool Fog computing Adopt unused PRB in adjacent cells Use transferred PRB from adjacent cells Rate is compensated Fig. 2 Function block diagram of the proposed scheme Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 5 of 15 to the fog device will be sent as UE is added or switched where n represents the total number of PRBs in the sys- in a small cell if resources are lacking. If there are no tem resource pool in its initial state, n denotes the total available idle resources in the system resource pool or in number of PRBs in the system resource pool before each adjacent small cells, then a PRB transfer process will be allocation, n n,and n − Y represents the remaining i i launched for the small cell. The adjacent small cells trans- PRBs in the system resource pool after allocating for small fer their PRBs to the small cell following certain rules, and cell s . I is an interference matrix of small cells. Any two the rate compensation is executed in adjacent small cells; cells whose signals interfere with each other are termed that is, the fog device finds available PRBs for the small “adjacent cells.” If s and s are adjacent cells, according to i l cells with rate loss. This process aims to achieve minimal the interference weight, I , I ∈ (0, 1]; otherwise, I = i,l l,i i,l PRB switching while facilitating user access. I = 0. A kind of demand function Y is proposed and l,i i showninformula (4b) for interference optimization and 4.1 Global resource allocation algorithm fairness in the whole network. Y means that the total PRB 4.1.1 Global resource allocation model in the system resource pool is shared equally by a small Based on the system model in Fig. 1, the cloud processing cell and its adjacent cells. In other words, the more adja- platform allocates a certain number of PRBs for each fog cent cells a small cell has, the more interference it causes, device, and the fog device has a PRB list of small cells. This and the Y is smaller. ∗ represents rounding down to process avoids interference to reduce user access delay the nearest integer. The constraint (4c) means that two and signaling overheads. The problem of global resource small cells with interference cannot be allocated to the allocation can be transformed into an optimization model, same PRB. which is aimed at the maximum number of remainder 4.1.2 Solution of global resource allocation model PRBs in the system resource pool after each allocation, as In this section, a global resource allocation optimization shown in formula (4). algorithm is proposed to solve the above optimization model. The algorithm is as follows: arg max n − Y , (4a) i=1 1. Each small cell scans the field strength and subject to ⎢ ⎥ forms an interference diagram of its neighbor ⎢ ⎥ ⎢ ⎥ cells. Meanwhile, the small cell reports the ⎢ ⎥ ⎢ ⎥ (4) n interference diagram to the cloud processing ⎢ ⎥ Y = (4b) ⎢ ⎥ platform. ⎢ ⎥ ⎣ 1 + I ⎦ i,l 2. The cloud processing platform generates the l=1 global interference graph shown in Fig. 3 and l=i ∀b ∈ BS , I = 1, ∃b ∈ / BS , (4c) forms the interference matrix. The vertex V of j i i,l j l SC1 SC2 SC3 SC8 SC7 SC5 SC6 SC4 Fig. 3 The global interference graph Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 6 of 15 the interference graph represents the small cell recover the PRB from the used PRB list of the powered-off in area G.The edge E of the interference graph small cell. If the network adds small cell s , the mathemati- stands for the conflicting relationship between cal optimization model aimed at the minimum amount of any two small cells that need to be avoided in PRB switching is interference. I and I indicates the weight of i,l l,i edge connecting small cells s and s . I depends i l i,l arg min T . (5a) on the SINR level in small cell s considering the i=1 interference of small cells s ,and I depends on l l,i subject to (5) the SINR level in small cell s considering the size BS ∝ Y . (5b) interference of small cell s . In other words, the size B ∝ size B . (5c) larger SINR is, the smaller the corresponding interference weight is. Here, we normalized the interference weights, I , I ∈[0,1]. Any two i,l l,i Here, the amount of PRB switching is defined as the vertexes with an edge cannot be allocated to the sum of the number of PRB changes used by the cell after same PRB. optimization. T stands for the extent of PRB switching 3. According to the formula (4b), the cloud caused by small cell s .size(BS ) represents the total num- processing platform calculates the number of ber of allocated PRBs in small cells after global allocation. allocated PRBs for each small cell in the set of S . size(B ) indicates the total number of PRBs in the sys- 4. The cloud processing platform allocates PRBs tem resource pool before partial resource optimization, for small cell s and chooses the PRB with the i while size(B ) is the total number of PRBs in the sys- smallest serial number in the system resource tem resource pool thereafter. The constraint (5b) indicates pool until the PRB number reaches Y , and the that the total number of practically allocated PRBs for chosen PRB has not been allocated to its small cell s isequaltoornearthe number of theoretically adjacent cells. allocated PRBs for small cell s . The constraint (5c) shows that the number of remaining PRBs in the system resource The pseudo code of algorithm 1 is as follows: pool is unchanged after resource optimization. To solve problem (5), the specific process of the resource allocation algorithm is as follows: Algorithm 1 Global allocation for small cells Require: c = 0, j = 1; //c represents the PRB number 1. A new small cell scans the field strength and allocated for s ; j stands for the index of PRB. i formats an interference diagram of neighbor Ensure: BS ; //BS represents the allocated PRB list of s . cells. Meanwhile, the new small cell reports the i i i 1: for i = 1; i <= m; i ++ do interference diagram to the cloud processing 2: while c <= Y do; //allocate PRB for s to meet its i platform. demand Y . i 2. The cloud processing platform updates the 3: if prb ∈ / BS and I = 1 then; // allocate PRB j i,l interference graph of adjacent cells according to for s different from its neighbors. the report on the relationship of the new small cell’s neighbors. 4: BS = BS + b ; i i 3. The cloud processing platform calculates the 5: c = c + 1; number of allocated PRB for new small cell 6: else using formula (4b). 7: j = j + 1; 4. BZ stands for the idle PRB list belonging to 8: end if i adjacent cells of small cell s .If b ∈ BZ and 9: end while i j s 10: end for b ∈ / B , the cloud processing platform allocates PRB b to small cell s . j i 5. If there are more than two adjacent cells with idle allocated PRB, the cloud processing 4.2 Partial resource optimization algorithm platform adjusts the PRB allocation of adjacent When the network topology changes, a partial resource cells according to fairness principles and tries its optimization algorithm is proposed, which is used to best to meet the demand of PRB allocation for reduce amount of PRB switching caused by twice global small cell s . Meanwhile, the cloud processing allocation, thus improving user satisfaction. When the platform deletes the corresponding PRB number of small cells in the network declines, such as allocated for small cell s from the idle PRB list when a small cell powers off, the system resource pool will of adjacent cells. Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 7 of 15 The fairness principles are as follows: set P(C ) to be the is unable to satisfy the rate demand of UE. This process probability that PRB b is capable of being borrowed, and may cause PRB switching in adjacent cells. In order to set P(D ) to be the probability that PRB switching is gen- guarantee minimal PRB switching in the whole network, erated in adjacent cells after borrowing PRB b .There is and to ensure the network throughput does not drop, a an adverse effect on subsequent resource allocation after mathematical optimization model is depicted as (11). borrowing some certain PRBs; thus, the cloud process- ing platform chooses the PRB with the smallest posterior arg min T (11a) (11) i=1 probability P(C |D ) to allocate to the new small cell. i i subject to O > O . (11b) P(Di|C ) downstream downstream Because P(C |Di) = and P(D ) does not depend i i P(D ) on C , the cloud processing platform selects the PRB with The system downlink throughput O in certain i downstream the smallest posterior probability to transfer to small cell transmission time t is calculated as s is shown in formula (6). i (t − t ) delay O = R · , (12) downstream k log P(C , D ) = log(P(D |C )) + logP(C ).(6) i i i i i k∈U log P(C , D ) is a type of Bayesian score. The prior distri- where t is a certain constant. O represents i i delay downstream bution for parameter θ is P(θ |C ). P(D |C ) in the formula the system downlink throughput before each resource c c i i i (6)isexpandedtoget theformula (7). coordination, while O represents the system downstream downlink throughput after each resource coordination. +∞ P(D |C ) = P(D |C , θ ) · P(θ |C )dθ.(7) The constraint (9b) means that the system downlink i i i i c c i c −∞ throughput will be increased after resource coordination. The first term of the integral in the formula (7)is To solve problem (11), a small cell resource coordination deconstructed to obtain the next formula. and scheduling algorithm (RCS) is proposed in fog device. N The specific implementation process of RCS algorithm is i,j,k P (D |C , θ ) = θ · i i c i,j,k as follows: i j k (8) 1. The fog device counts the small cells with new i,j,k θ , PRB demands for the formation of the set i,j,k i j k S = s , s , ... . 1 2 2. If the remaining PRB of the system resource where θ = P x |a is the conditional probability i,j,k i i pool can meet the new PRB demands, the fog where x is equal to k in the case such that the parent node device allocates the PRB to small cell s and of the ith node x is j. Therefore, the prior distribution deletes the PRB from the system resource pool. for parameter θ canbebrokendownintothe following Then, the process jumps to step 7, where the formula: resource pool will recover the PRB when the user communication is completed; otherwise, P(θ |C ) = P θ · P θ .(9) c i i,j,k i,j,k the fog device solves unmet PRB demands of i j i j small cell s in the next step. Apply the formula (8)and (9)totheformula(7). 3. If the adjacent cells have idle PRBs, the fog +∞ device distributes these PRBs to the small cell s . i,j,k ◦ ◦ ◦ P(D |C ) = θ ·P θ dθ i i If the PRB requirements have been met, the i,j,k i,j,k i,j,k −∞ i j k process proceeds to step 7. Otherwise, the process proceeds to step 4. (10) 4. If small cell s never initiates the PRB transfer Assuming that the initial variables N and the transfer i,j,k process but the transfer will improve the system variables N have been selected and the observed data throughput, the fog device counts the PRB i,j,k are complete, the problem can be solved using a heuristic adjacent cells used and calculates the PRB value algorithm (e.g., the greedy algorithm, genetic algorithm, using the formula shown in (13). The PRB with tabu search algorithm, or ant colony algorithm). the maximum value is prohibited from transferring. The fog device selects the PRB with 4.3 Resource coordination and scheduling algorithm the minimum value to transfer to meet the In the system, each small cell has an idle allocated PRB demand of small cell s . In particular, if multiple list and a used PRB list. When a new UE or switching UE PRBs have equal values, the fog device chooses initiates an access request to the serving small cell, the the PRB with the smaller index to transfer; smallcellneedstoget thereusablePRB if itsidlePRB otherwise, the process proceeds to step 7. Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 8 of 15 j,l 1. The cloud processing platform allocates PRB for the V = . (13) size(BS ) powered-on small cells based on the global resource s ∈Z allocation algorithm. 2. After small cell s powers on, the fog device borrows Here, V stands for the value of PRB b in j j PRB b and b from the idle PRB list in adjacent cells 3 4 adjacent cell s ,and Z represents the adjacent to allocate for small cell s based on the partial cell set of small cell s . The fog device forms the resource optimization algorithm. In this case, b and PRBs set BS by sorting the PRB in ascending b remain in system resource pool. order based on the use value in adjacent cells. 3. Small cell s has five new PRB demands and uses the PRB’s use value depends on the actual rate allocated b and b . The new PRB demand of s has 3 4 8 obtained by UE who allocated this PRB in small not been met. At this time, a system resource cell. w stands for the index of PRB b in BS . j,l j allocation conflict occurs. The fog device finds available PRBs using a resource coordination and 5. If the PRB small cell s needs have been met, the scheduling algorithm. The specific steps of this small cell s exits the resource scheduling process are as follows: process and does not participate in subsequent coordination, the process proceeds to step 6. (a) The fog device allocates PRB b and b in 9 10 Otherwise, the process proceeds to step 7. the system resource pool to small cell s and 6. The fog device calculates the rate loss of a small deletes the corresponding PRB from the cell after transfer based on Eq. (14)and sorts resource pool. The new PRB demand of small cells in descending order according to the small cell s has not been satisfied, and rate loss. The ordered results will be sorted into adjacent cells have no idle PRB. The fog the set S , prioritizing the resource scheduling device launches the transfer process for process to those small cells whose new PRB small cell s and uses the formula (13)to demands have not yet been met. calculate the value of PRB used in adjacent cells. The PRB with the maximum value is b s ∈TS R prohibited from transferring. The fog device l,k,j j=1 L = (14) selects the PRB b with the minimum value b s ∈BS l l l,k,j to transfer to small cell s . The PRB demands j=1 8 of small cell s have been met, and small cell In this equation, L represents the rate loss of s exits the resource scheduling process and adjacent cell s , TS denotes the set of does not participate in subsequent l l coordination. The rate loss caused by the transferred PRBs in adjacent cell s ,and BS transfer of small cell s is , and that caused stands for the PRB list of adjacent cell s before l 1 the transfer. by the transfer of small cell s is . 7. If the PRB demands in the set S have been (b) The fog device launches the rate solved, the algorithm ends; otherwise, the compensation for small cell s ,and smallcell process proceeds to step 2. s uses its own allocated PRB b to meet 1 7 demands. Then, the fog device initiates the 5 Resource scheduling example rate compensation for small cell s and This section introduces a PRB scheduling example of allocates the unused PRB b in adjacent cells small cells to better explain the implementation pro- to small cell s . Once the resource cess of the proposed scheme. Small cells belong to the requirements of all small cells in the system same fog device as they are deployed in the adjacent have been met, the algorithm ends. office of the same building. Seven small cells s , s , ..., s 1 2 7 Assume there is no UE access to small cell s for an power on, while s is in a shutdown state. After a cer- extended period of time, and the allocated PRBs of small tain period of time, s powers on and can no longer cell s are all borrowed to its adjacent cells. Conversely, accommodate a new access request. In this situation, a 8 when there are new accessed users in small cell s ,the sys- system resource allocation conflict occurs. As is shown in 8 tem will initiate the PRB transfer process. As shown in Fig. 4, in actual network operation, there is an example of Fig. 5, there are two new PRB demands in the powered- resource scheduling in managed area of a fog device. In on small cell s . In order to allocate PRB for new users this example, the “j”standsfor b having been used; “j” 8 of small cell s , the PRB transfer will occur among the indicates PRB b has not been used. The scheduling case 8 other small cells. As a result, small cell s will get enough is primarily divided into four processes: 8 Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 9 of 15 SC 1 SC 1 a) allocate PRB for b) SC8 is on, PRB pool PRB: PRB: small cells. allocate PRB SC 2 SC 2 9,10 3,4,5,6,7 5,6,7 for SC8. PRB: PRB: SC 3 SC 3 6,7,5 5,6,7 SC 8 PRB:1,2 SC 8 PRB:1,2 off SC 7 PRB:3,4 SC 7 SC 4 PRB: SC 4 PRB: PRB: 3,4 PRB: SC 5 3,4 SC 5 5,6,7 5,6,7 PRB: PRB: c1) Five new PRB 1,2,5 1,2,5 demands in SC8. SC 6 SC 6 PRB: PRB: 3,4,6,7,8 SC 1 3,4,6,7,8 PRB pool PRB:6,7 SC 2 null SC 1 (loss) PRB:7,6 SC 3 SC 8 PRB:6,7 SC 2 c2) Compensation ,5 PRB:1,2 SC 8 PRB:3,4 SC 3 PRB:6,7 for SC1 and SC4. PRB:1,2 PRB:3,4 ,9,10,5(e ,5 SC 7 SC 4 ,9,10,5(e xit) PRB:3,4 SC 4 SC 7 PRB:6,7 SC 5 xit) SC 5 PRB:8,6 PRB:3,4 (loss) PRB:1,2 PRB:1,2 ,7 ,5 ,5 SC 6 SC 6 PRB:3,4 PRB:3,4 ,6,7,8 ,6,7,8 Fig. 4 Resource scheduling example PRB for its new users, while the used PRB from other graph coloring algorithm. The global resource allocation small cells will need to make additional PRB switches. The can reduce the delay of PRB allocation for new user access. system overhead and transmission delay caused by PRB In this case, small cell s is the worst case, and it does switching should be minimized as much as possible. Here, not fully leverage the advantages of allocation. However, we give a scheduling case comparison of the clustering- the algorithm still reduces the amount of PRB switch- based interference coordination algorithm [16]and the ing caused by resource conflicts and lowers the system graph coloring algorithm [27] to our algorithm based on overheads. the allocation in Fig. 5 to analyze and contrast transmis- sion delay and signaling overheads. The clustering-based 6 Simulation interference coordination algorithm was described in the 6.1 Scenario setting introduction, and the graph coloring algorithm is a type of In the simulation scenario, 64 small cells were deployed greedy algorithm. The latter is a heuristic method of solv- within a macro cell coverage range. The macro and small ing the inter-cell interference problem and involves two cells used differ-frequency. UE randomly arrived at a small main stages: (i) the formation of interference graph and (ii) cell at the arrival rate of 30 user/s, and its transmission rate the resource allocation solution. The resource allocation demands range from 1 to 1000 kb/s. There were 25 total solution means that each nodes must be allocated a pre- PRBs in the system. The specific simulation parameter will set number of colors, and no two adjacent nodes may have be written as shown in Tables 1 and 2. To determine the any colors in common. The objective is to complete this performance of the proposed algorithm, we compared it using the fewest possible number of colors. PRB switching to two other algorithms with regard to the following: sys- occurs three times in the proposed algorithm, six times tem throughput, average user SINR, user satisfaction, and in the clustering-based algorithm, and nine times in the signaling overheads. Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 10 of 15 PRB:1,2 ,10 PRB:4,7 ,8 PRB:5,6 PRB:3,4 Clustering-based algorithm. SC 4 PRB:3,4 PRB:4,7 The allocation of PRB at a ,8,9 PRB:4,5 certain time. ,6 PRB:1,2 PRB:7,3 ,10 PRB:1 ,8 PRB:5,6 PRB: null Graph coloring PRB:3,4 PRB:2,3 algorithm. PRB:1,2 ,4,7 ,3 PRB:4,9 PRB:1,7 ,10 ,8 The proposed PRB:5,6 PRB:1 PRB:3,4 algorithm. PRB:3 PRB:1,7 ,4 ,8,9 PRB:1,5 ,6 PRB:5,2 ,10(exit) SC 3 PRB:2 PRB:7,3 PRB:9,6 ,8 PRB:1,3 (exit) PRB:2,8 PRB:3,4 ,4,7 PRB:4,9 ,10 PRB:1 Fig. 5 Comparison scheduling case with other two algorithms 7 Results and discussion 7.1 Analysis of throughput and SINR Table 1 System parameter The simulation curves in this study are zigzagged because Parameters Value the user arrival rate (30 user/s) caused the performance index to change significantly. As shown in Fig. 6,when Simulation area size 600m*600m the user number is less than 210, the system resources Bandwidth of network 5MHz/2GHz are abundant and can meet users’ demands , and the Antenna gain of small cell 5dB system throughput increases overall. The system through- Maximum transmit power of BS 24dBm put reaches a maximum about 121 Mbps when the user Radius of the small cell 40m number is equal to 210. The proposed algorithm satis- Maximum user transmit power 23dBm fied the resource demands of users in serving small cells by borrowing resources from adjacent cells. Some users Handover delay 20ms were assigned the PRB with a low rate and high reusabil- White noise −174dBm/Hz ity, which scarified the benefit for users assigned this Shadow fading standard deviation 3dB PRB, and clearly improved the throughput of a new user Number of User[initial value: step: final value] [30:30:750] obtaining this PRB; therefore, the system throughput is Buffer model Full buffer improved significantly. When the user number gradually Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 11 of 15 Table 2 Table of notations increases to above 210 and the overall trend becomes Symbol Meaning saturated, the algorithm is convergent. New users can- not access the network when there are more than 210 S = {s , ..., s } Set of small cells in the whole network, m ∈ N ,and 1 m N represents the set of positive integers. users. The throughput of the other two algorithms are B = {b , ...b } Set of PRBs in the system resource pool, n ∈ N . lower than the proposed algorithm, and the proposed 1 n algorithm tended to be saturated more quickly; hence, the U = {u , ..., u } Set of users (UE) of a small cell, o ∈ N . 1 o proposed algorithm clearly achieved the desired results of R The rate user u obtained based on PRB b in small i,j,k k j improving the system throughput. cell s . Through the coordination of small cells, this algorithm W Bandwidth of per PRBs. demonstrated better performance compared to the other Sinr SINR u receivedbasedonPRBs b in small cell s . i,j,k k j i two algorithms for mitigating inter-cell interference and g Channel gain. i,j,k improving SINR. As illustrated in Fig. 7, when the num- p Transmitted power of small cell. ber of users increased, the other two algorithms showed n Noise power. a sharp increase in interference, while the proposed algo- rithms shows a slow increase and its SINR tended to BS = {b s , b s , ..., } The set of available PRBs in small cell s . i 1 i 2 i i converge quickly. R A constant to indicate the rate demand of user u . k k n Total number of PRB in the system resource pool in 7.2 Analysis of user satisfaction initial state. User satisfaction is a comprehensive evaluation of users’ n Total number of PRBs in the system resource pool successful access and continuous communication. The before each pre-allocation, n n. user is unable to access to the network until the small cell i {i|1 ≤ i m}, m is the total number of small cells. meets its rate demand, and the probability of successful n − Y The remainder PRBs in the system resource pool access is P . With increasing user numbers, the PRB access after pre-allocation for s . used in small cells will be transferred to the new users I An interference matrix of small cells. to ensure throughput maximization when the wireless Y The number of allocated PRBs in small cell s . i i resource is scarce, which will cause some communicating S = s , ..., s The set of small cells in the whole network in 1 m users of being dropped. The probability to be dropped is descending order based on the degree of vertex in P . drop interference graph. User satisfaction with the network services measures BS The set of PRBs in small cell s after global pre- i the network’s performance. Here, the user satisfaction is allocation. defined as P [28]: B Total number of PRBs in the system resource pool before resource optimization. P = 0.1 × P + 0.9 × (1 − P ). (15) access drop B Total number of PRBs in the system resource pool after resource optimization. From the user’s experience with network services, the BZ The idle PRB list belonging to adjacent cells of small user prefers to be denied access rather than to be dropped. cell s . 1 − P stands for the probability of not being dropped; drop Z The adjacent cell set of small cell s . i therefore, in this formula, the weight factor of success- O System downlink throughput. downstream ful access is 0.1, and the weight factor of not being O System downlink throughput after each resource dropped is 0.9. downstream coordination. As is shown in Fig. 8,withanincreaseinusernumbers, t Average processing time caused by PRB switching. user satisfaction starts to decline when the user number delay reaches 150. User satisfaction decreases slowly and gradu- V The value of PRB b . j j ally trends plateaus when the user number is greater than S = s , s , ... The set of small cells with new PRB demands. 1 2 150. Compared to the other two algorithms, the proposed BS The set of PRBs in ascending order based on the algorithmobtains abetterusersatisfaction. Becausethe value in the adjacent cell s of small cell s . l l algorithm solves the small cell resource allocation con- O System downlink output after each resource coordi- downstream flict harmonically in the local area, it reduces the PRB nation. switching and users’ outage probability. w The order of PRB b in BS . j,l j L The rate loss of the adjacent cell s of small cell s . l l i 7.3 Analysis of signaling overheads TS The set of transferred PRBs in small cell s . l l Figure 9 offers a simple example to indicate that the proposed algorithm has less signaling than the other algo- rithms on the basis of two situations. Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 12 of 15 Fig. 6 Comparison of system throughput. Users include those with access and those denied access In Fig. 9a, the PRB of the small cell is sufficient. In before. In other words, the PRB switching is the main initial state, the “accessed UE” accesses the small cell, reason for the increase in signaling overhead. which needs two signaling (the solid line arrows). In gen- In Fig. 9b, when the “new UE” requests to access the eral, the proposed algorithm causes less access delay for small cell β, the proposed algorithm will execute the two signaling than the other two algorithms because the resource coordination and scheduling process if the PRB global resource allocation for the small cell managed by of the small cell is deficient. If the small cell α has idle fog devices. When the “new UE” accesses the small cell, available PRB, it can borrow PRB to small cell β.Other- there are also two signaling between new UE and the wise, the small cell can transfer the PRB with the smallest small cell. The proposed algorithm has fewer instances of used value to the small cell β, which may lead to signal- PRB switching than the other two algorithms. There are ing for PRB switching; thus, there are two extra signaling two extra signaling (the dotted arrows) in the other two between small cell α and the “accessed UE.” The other two algorithms because they require global resource allocation algorithms require the global resource allocation for all again once the new UE requests access, and the “accessed the UE within the network, so this process may cause two UE” in small cell may be assigned to a different PRB than signaling to each “accessed UE.” Fig. 7 Comparison of average SINR. Users include those with access and those denied access Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 13 of 15 Fig. 8 Comparison of user satisfaction. Users include those with access and those denied access Assume that the signaling overhead per PRB switch user number increases. When the user number reaches a is equal to 20 ms. As shown in Fig. 10, when the user certain value, the signaling overhead gradually decreases number is less than 150 (i.e., the PRB is sufficient), the due to the number of access users becoming saturated, signaling overhead of the proposed algorithm remains at and the system capacity reaches its limit. Meanwhile, sig- zero and the other two algorithms gradually increase as naling overhead of the proposed algorithm is less than the user number increases. When the number of users the other two algorithms, proving PRB switching times is larger than 150 (i.e., the PRB is deficient), the sig- in this algorithm is less frequent than in the other two naling overhead of all three algorithms increases as the algorithms. Signaling Accessed UE Accessed UE New UE New UE The proposed algorithm. The other two algorithms. Accessed UE Accessed UE Accessed UE Accessed UE New UE New UE Accessed UE Accessed UE The proposed algorithm. The other two algorithms. Fig. 9 Comparison of signaling in access process. a The PRB of small cell is sufficient. b The PRB of small cell is deficient Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 14 of 15 Fig. 10 Comparison of signaling overheads. Users include those with access and those denied access 8Conclusions Funding The research in this paper and its publication are supported by the National The problems of interference and system overhead in Natural Science Foundation of China (Grant No. 61731012) and the National ultra-dense small cells must be resolved. They depend on High Technology Research and Development Program of China (Grant No. the efficient and rational allocation of radio resources. 2015AA01A705). This study presents a novel radio resource coordina- Authors’ contributions tion and scheduling scheme based on cloud infrastruc- YT proposed the research ideas and directed the writing of the paper. ZC is the main author of this paper; she conducted the experiment and analyzed it ture in an ultra-dense small cell network. In light of with the help of SZ. Other authors promoted the quality of the entire Cloud-RAN’s improved computational ability, the global manuscript. All authors read and approved the final manuscript. resource allocation and partial resource optimization Competing interests with heavy computing tasks can be run on the cloud The authors declare that they have no competing interests. processing platform, considering the resource utiliza- Publisher’s Note tion and the fairness of resource allocation between Springer Nature remains neutral with regard to jurisdictional claims in small cells in the entire network. Meanwhile, because published maps and institutional affiliations. fog computing has a lower transmission delay, the Author details resource coordination and scheduling algorithm are pro- Department of Communication Engineering, Xiamen University, Xiamen posed for generally changeable small cells via the fog 361005, China. Department of Computer and Information Sciences, Temple University, Philadelphia, USA. Department of Electrical and Computer device, thereby avoiding excessive signaling overheads Engineering, University of Idaho, Moscow, ID, USA. caused by using cloud computing. The simulation results Received: 31 March 2017 Accepted: 15 May 2018 show that the proposed scheme improves user satisfac- tion and downlink throughput and reduces system sig- naling overheads through the resource coordination of References small cells. 1. S Chen, F Qin, B Hu, X Li, Z Chen, User-centric ultra-dense networks for 5G: challenges, methodologies, and directions. IEEE Wireless Commun. 23(2), Acknowledgements 78–85 (2016) This research was supported by the National Natural Science Foundation of 2. O Semiari, W Saad, S Valentin, M Bennis, B Maham, in Acoustics, Speech China (Grant No. 61731012) and the National High Technology Research and and Signal Processing (ICASSP), 2014 IEEE International Conference On. Development Program of China (Grant No. 2015AA01A705). An earlier version Matching theory for priority-based cell association in the downlink of of this paper was presented at the 2017 12th International Conference on wireless small cell networks (IEEE, Florence, 2014), pp. 444–448 Computer Science and Education (ICCSE). Special thanks are dedicated to 3. C Yang, J Li, M Guizani, Cooperation for spectral and energy efficiency in Professor Weihua Zhuang’s guidance, from the Department of Electrical and ultra-dense small cell networks. IEEE Wireless Commun. 23(1), 64–71 Computer Engineering at the University of Waterloo, Canada. (2016) Chen et al. EURASIP Journal on Wireless Communications and Networking (2018) 2018:137 Page 15 of 15 4. I Chih-Lin, J Huang, R Duan, C Cui, JX Jiang, L Li, Recent progress on C-RAN centralization and cloudification. IEEE Access. 2, 1030–1039 (2014) 5. JG Andrews, S Buzzi, W Choi, SV Hanly, A Lozano, AC Soong, JC Zhang, What will 5G be? IEEE J. Selected Areas Commun. 32(6), 1065–1082 (2014) 6. S Zou, F Yang, Y Tang, L Xiao, The resource mapping algorithm of wireless virtualized networks for saving energy in ultradense small cells. Mob. Inf. Syst. 2015, 1–14 (2015) 7. S Sarkar, S Chatterjee, S Misra, Assessment of the suitability of fog computing in the context of internet of things. IEEE Trans. Cloud Comput. 6(1), 46–59 (2015) 8. M Bennis, SM Perlaza, P Blasco, Z Han, HV Poor, Self-organization in small cell networks: A reinforcement learning approach. IEEE Trans. Wireless Commun. 12(7), 3202–3212 (2013) 9. A Ghosh, R Ratasuk, B Mondal, N Mangalvedhe, T Thomas, LTE-advanced: next-generation wireless broadband technology [invited paper]. IEEE Wireless Commun. 17(3), 10–22 (2010) 10. S Bassoy, H Farooq, MA Imran, A Imran, Coordinated multi-point clustering schemes: a survey. IEEE Commun. Surv. Tutorials. 19(2), 743–763 (2017) 11. A Damnjanovic, J Montojo, Y Wei, T Ji, T Luo, M Vajapeyam, T Yoo, O Song, D Malladi, A survey on 3GPP heterogeneous networks. IEEE Wireless Commun. 18(3), 10–21 (2011) 12. Y Dong, Z Chen, P Fan, KB Letaief, Mobility-aware uplink interference model for 5G heterogeneous networks. IEEE Trans. Wireless Commun. 15(3), 2231–2244 (2016) 13. J Zheng, Y Cai, A Anpalagan, A stochastic game-theoretic approach for interference mitigation in small cell networks. IEEE Commun. Lett. 19(2), 251–254 (2015) 14. S Lin, H Tian, in Wireless Communications and Networking Conference (WCNC), 2013 IEEE. Clustering based interference management for QoS guarantees in OFDMA femtocell (IEEE, Shanghai, 2013), pp. 649–654 15. Y Zhang, S Wang, J Guo, in Communications in China (ICCC), 2015 IEEE/CIC International Conference On. Clustering-based interference management in densely deployed femtocell networks (EEE, Shenzhen, 2015), pp. 1–6 16. A Hatoum, R Langar, N Aitsaadi, R Boutaba, G Pujolle, Cluster-based resource management in OFDMA femtocell networks with QoS guarantees. IEEE Trans. Veh. Technol. 63(5), 2378–2391 (2014) 17. A Abdelnasser, E Hossain, DI Kim, Clustering and resource allocation for dense femtocells in a two-tier cellular OFDMA network. IEEE Trans. Wireless Commun. 13(3), 1628–1641 (2014) 18. T Lotfollahzadeh, S Kabiri, H Kalbkhani, MG Shayesteh, Femtocell base station clustering and logistic smooth transition autoregressive-based predicted signal-to-interference-plus-noise ratio for performance improvement of two-tier macro/femtocell networks. IET Signal Process. 10(1), 1–11 (2016) 19. Z Sun, R Liu, W Wang, Joint time-frequency domain cyclostationarity- based approach to blind estimation of OFDM transmission parameters. EURASIP J. Wireless Commun. Netw. 2013(1), 1–8 (2013) 20. B Han, W Wang, Y Li, M Peng, Investigation of interference margin for the co-existence of macrocell and femtocell in orthogonal frequency division multiple access systems. IEEE Syst. J. 7(1), 59–67 (2013) 21. TA Weiss, FK Jondral, Spectrum pooling: an innovative strategy for the enhancement of spectrum efficiency. IEEE Commun. Mag. 42(3), 8–14 (2004) 22. J Holdren, E Lander, Realizing the full potential of government-held spectrum to spur economic growth. Tech. Rep. (2012) 23. Y Li, Z Feng, S Chen, Y Chen, D Xu, P Zhang, Q Zhang, Radio resource management for public femtocell networks. EURASIP J. Wireless Commun. Netw. 2011(1), 1–16 (2011) 24. Y Li, Z Feng, D Xu, Q Zhang, H Tian, Optimisation approach for femtocell networks using coordinated multipoint transmission technique. Electron. Lett. 47(24), 1348–1349 (2011) 25. Z Chen, Y Tang, in Computer Science and Education (ICCSE), 2017 12th International Conference On. A resource collaboration scheduling scheme in ultra-dense small cells (IEEE, Houston, 2017), pp. 401–405 26. H Aissi, C Bazgan, D Vanderpooten, Complexity of the min-max and min-max regret assignment problems. Oper. Res. Lett. 33(6), 634–640 (2005) 27. A Mehrotra, MA Trick, A branch-and-price approach for graph multi-coloring. Oper. Res. Comput. Sci. Interfaces. 37, 15–29 (2010) 28. Y Wang, P Zhang, Radio resource management. (Beijing University of Posts and Telecommunications Press, China, 2005), p. 21
EURASIP Journal on Wireless Communications and Networking – Springer Journals
Published: May 31, 2018
Access the full text.
Sign up today, get DeepDyve free for 14 days.