Angle⁃Based Based Interference⁃Aware Aware Routing Algorithm for hm for Multicas

Q ian Xu,Pinyi Ren,Q inghe Du,Gang W u,Q iang Li,and Li Sun(.Department of Information and Communications Engineering，Xi′an Jiaotong University，Xi′an 70049，China;.Microelectronics Institute Algorithm Design Department，ZTE Corporation，Shenzhen 58057，China)

Angle⁃Based Based Interference⁃Aware Aware Routing Algorithm for hm for MulticastoverW ireless eless D D22DNetworks works

Q ian Xu1,Pinyi Ren1,Q inghe Du1,Gang W u2,
Q iang Li2,and Li Sun1
(1.Department of Information and Communications Engineering，Xi′an Jiaotong University，Xi′an 710049，China;
2.Microelectronics Institute Algorithm Design Department，ZTE Corporation，Shenzhen 518057，China)

Wireless device⁃to⁃device(D2D)communications sharing the spectrum of cellular networks is important for improving spec⁃trum efficiency.Furthermore，introducingmulticast and multi⁃hop communications to D2D networks can expand D2D ser⁃vice functions.In this paper，we propose an angle⁃based inter⁃ference⁃aware routing algorithm for D2D multicast communica⁃tions.This algorithm reuses the uplink cellular spectrum.Our proposed algorithm aims to reduce the outage probability and minimize the average hop count over all multicast destina⁃tions(i.e.，multicast receivers)，while limiting interference to cellular users to a tolerable level.In particular，our algorithm integrates two design principles for hop⁃by⁃hop route selec⁃tion.First，weminimize the distance ratio of the candidate⁃to⁃destination link to the candidate⁃to⁃base⁃station link，such that the selected route advances closer to a subset of multi⁃cast receivers.Second，we design the angle⁃threshold based merging strategy to divide multicast receivers into subsets with geographically close destinations.By applying the two principles for selection of each hop and further deriving an adaptive power⁃allocation strategy，the message can be more efficiently delivered to destinations with fewer branches when constructing the multicast tree.This means fewer duplicated data transmissions.Analyses and simulations are presented to show the impact of system parameters on the routing perfor⁃mances.Simulation results also demonstrate the superiority of our algorithm over baseline schemes in terms of outage proba⁃bility and average hop count.

device⁃to⁃device communications;multicast;interference⁃aware routing;cellular networks

1 Introduction

With the popularity of applications such as so⁃cial networks and local broadcast，which in⁃volve direct interaction between end users/ nodes，wireless device⁃to⁃device(D2D)com⁃munication hasaroused great interest in the research communi⁃ty[1]-[10].Wireless D2D communication reusing cellular net⁃work spectrum have shown great potential to relieve spectrum scarcity issues.Typically，D2D communications employ a non⁃orthogonal spectrum sharing approach[3]-[10]，which needs specifically designed sharing strategies to control the cross⁃in⁃terferencesbetween users.

Recently，research on coexistence of cellular and D2D net⁃works hasmainly addressed interference management strate⁃gies[3]，[4]，transmissionmode selection[5]，power control[6]，routing design[7]，resources allocations[8]，[9]，and network coding[10].However，the majority of research on D2D net⁃works has focused on one⁃hop point⁃to⁃point transmission be⁃tween device nodes，which limits the functions and applica⁃tions of D2D networks.First，D2D users need to lower trans⁃mission power to avoid causing intolerable interference to cel⁃lular users，and the transmission range is limited.Therefore，multi⁃hop D2D communications are highly desirable.Second，multiple users often require the same content，such as software download，online gaming，and video streaming.Unicast⁃based approaches for such services would waste considerable spec⁃trum.Thus，supportingmulticast functions in D2D networks is important to better utilize precious wireless resources.To ad⁃dress these issues，we concentrate on the routing design for multicast transmissionsover D2D networks.

Much research effort has been dedicated to multicast rout⁃ing over diversewireless networks[11]-[19].However，we still meetnew challenges introduced by the unique featuresofwire⁃less D2D networks.In particular，the routing scheme needs to be aware of interference caused to cellular users.Cellular in⁃terferences to D2D links also significantly affect route⁃selec⁃tion strategies.These interferences are called inter⁃network in⁃terferences.Moreover，as themulticast receiversare geographi⁃cally independent，themulticast treemightexpand withmulti⁃ple branches.As a consequence，intra⁃network interferences often exist across differentmulticast branches.Joint handling of these two types of interferences for efficientmulticast D2D routing remains an open problem that has not been thoroughly studied.

Aiming atsupporting efficientmulti⁃hopmulticastover D2D networks，we propose an angle⁃based interference⁃aware rout⁃ing algorithm formulticast D2D transmissions.Our proposed algorithm aims at lowering the outage probability for routing and minimizing the average hop count.Specifically，we select the route foreach hop by applying the distance ratiominimiza⁃tion and angle⁃threshold basedmerging principles.We design the power allocation strategies to the selected routers for eachhop to notonly avoid causing intolerable interference to cellu⁃lar users but also lower the outage probability.Simulation re⁃sults show that our proposed routing algorithm outperforms the baseline schemes in terms of outage probability and average hop count.

The remainder of this paper is organized as follows.Section 2 reviews the related work.Section 3 presents the system mod⁃el.Section 4 proposesour angle⁃based interference⁃aware rout⁃ing algorithm.Section 5 conducts analyzes and discussions on our proposed routing algorithm.Section 6 evaluates the perfor⁃manceson our proposed algorithm through simulations.The pa⁃per concludeswith section 7.

2 Related W ork

Recent research on wireless D2D networks has not properly addressed the multicast routing issue.In contrast，multicast routing schemes have been proposed in other types of net⁃works，which provide useful information and valuable referenc⁃es for our design in D2D networks.We first review existing re⁃search outcomes onmulticast routing in the cognitive networks [13]-[15]as cognitive and D2D networks reveal similar fea⁃tures in interference control and protection for prior users. Then we discuss somemulticast routing schemes in other net⁃works，such asmobilead⁃hoc networksand wirelesssensornet⁃works[16]-[19].

The authors of[13]jointly considered scheduling and rout⁃ing.They formulated the problem by a mixed⁃integer linear program and developed a polynomial⁃timealgorithm according⁃ly to identify a near optimal solution.In[14]，an on⁃demand multicast routing and channel⁃allocation algorithm called OM⁃RA was proposed.The OMRA algorithm senta signaling pack⁃et to build themulticast tree.The authors of[15]investigated the routing problem，where directional antenna is applied in contrast to thegenerally used omnidirectionalantennas.

Multicast routing forwirelesssensornetworkshasbeen stud⁃ied in[16]，[17]and[18].The authors in[16]firstbuilta heu⁃ristic virtual Euclidean Steinermulticast tree by using a specif⁃ically designed metric termed reduction ratio.Then，each for⁃warding node divides destinations into subsets based on the virtual tree and selects respective next hop for each subset. The authors of[17]proposed a geographic multicast routing (GMR)algorithm forwireless sensor networks.The GMR algo⁃rithm makes a tradeoff between multicast tree cost and effec⁃tiveness of data distribution.The authors of[18]proposed a routing scheme that avoids getting close to the interfering source.In particular，each forwarding node uses theminimum required transmission power to represent the interference strength level and thenmakes a routing decision.The authors of[19]proposed a position⁃based multicast(PBM)routing scheme formobile ad⁃hoc networks.The proposed PBMalgo⁃rithm does not need to build or expand the data distribution structure such asmulticast tree and/ormesh grid.Next⁃hop se⁃ lection is only based on the positions of the forwarding node，associated neighboring nodes，and destinations.All these algo⁃rithms have their respectivemerits but cannot be directly ap⁃plied to the D2D networksbecause of the unique features in in⁃terferencemanagement.In the following sections，discuss the design of an angle⁃based interference⁃awaremulticast routing scheme.

3 System Model

3.1 System Descriptions

We consider a scenariowhere the D2D network and cellular network coexist(Fig.1).For the cellular network，we consider one base station and focus on one cellular user′s uplink trans⁃mission with bandwidth equal to B Hz.The D2D network con⁃tains U D2D device nodes(indexed by 1，2，...，U)，wherewe fur⁃ther define the D2D node set byℵ≜{} 1，2，...，U.All D2D nodes can establish direct communications linkswith each oth⁃er by reusing the cellularuser′suplink spectrum.

In the D2D network，there is one multicast session where one source node Msr(Msr∈ℵ)attempts tomulticast a com⁃mon message to m destinations(multicast receivers) M1，M2，...Mm∈ℵthrough multi⁃hop connections.As shown in Fig.1，other than source and destination nodes，the rest of D2D nodes can work as the relay nodes(also called forwarding nodes)to help forward themulticastmessage towards destina⁃tions.The destination nodes themselves can function as the re⁃lay nodes.Following this setup，the source node，destination nodes，and relay nodes together form a multicast tree.As in Fig.1，amulticast treeoften expandsmultiplebranches.There⁃fore，theremight bemultiple D2D nodes transmitting packets simultaneously，which causes intra⁃network interference be⁃tween simultaneous transmissions of different D2D links.On the other hand，D2D links reuse the spectrum of cellular users，and there is interference between the D2D links and cellular links.This results in inter⁃network interference.Formulticast routing over D2D networks，we need to take both the inter⁃andintra⁃network interferences intoaccount.

▲Figure1.System model for D2Dmulticastcommunicationscoexisting with a cellular network.

3.2 Multicast Tree

Following the system model introduced in section 3.1，we next present several rigorous definitions to precisely describe themulticast treestructure.

Definition 1(D2D Multicast tree):A D2Dmulticast treeκ is a tree structure indicating the paths for packets to be deliv⁃ered tomulticast destinations.Generally，a tree comprises root node，internal nodes，and leaf nodes.For multicast tree，the rootnode is the source node.All leaf nodes in the tree are des⁃tination nodes for multicast;however，destination nodesmay not be leaf nodes because they can also function as internal nodes.A direct connection between two nodes is called a hop or an edge，where a hop connecting two nodes a and b，and a，b∈ℵ，is written as(a，b).Furthermore，for the hop(a，b)，a and b are termed transmitting and receiving node，respectively.

Definition 2(Level):The levelofa node in themulticast tree is defined by one plus the number of edges from the source (root)to this node.Accordingly，the source node Msris at the first level.

Definition 3(SL⁃path):Source⁃to⁃leaf path(SL⁃path)in a multicast tree represents a sequence of nodes and hops(edges) connecting the source node to a leaf node.Assume that there are total Y SL⁃paths in amulticast tree and then we use Wy，y=1，2，..，Y，to denote the setofall edges on y th SL⁃path.We then have Y≤m.We also define Qy，y=1，2，..，Y，as the set ofnodeson the y th SL⁃path.

Definition 4(x th hop):We define the x th hop belonging to an SL path as the hop connecting thenode at the x th level(also called the x th node of this SL path)and thatat the(x+1)th lev⁃el.The x th hop′s receiving node is the(x+1)th hop′s transmit⁃ting node.Furthermore，we use Hxyto denote the x th hop of y th SL⁃path and Txyand Rxyto denote the transmitting node and receiving node of hop Hxy，respectively，where y=1，2，…，Y and x=1，2，…，|Wy|.Here Y is the totalnumber of SL paths，and||Wyis the number of hops on the y th SL path，both ofwhich depend on how themulticast tree is built，The cardinality of the set isdenoted|·|.

Definition 5(x th hop set):We define the x th hop set Hxas the set of the x th hop belonging to each SL path，i.e.，In addition，we use Txand Rxto denote the set of transmitting nodes and receiving nodes，respectively，of hops in Hx.

Fig.2 shows amulticast tree example，with a source node Msr=a，four destination nodes M1=d，M2=e，M3=f，and M4=g，and two other D2D nodes b and c，where a through g are different elements in the D2D node setℵ.As shown in Fig.2，there are three leafnodes d，e，and g，and three internal nodes b，c，and f，where f is also a destination node.We can see from Fig.2 that there are three SL⁃paths，denoted by W1，W2，and W3，where W1=and W3={(a，c)，(c，f)，(f，g)}.Using the first SL path as a typ⁃and receiving nodes asH23=(c，f)，forming the 2nd hop setfor all SL⁃paths.In addition，we get

▲Figure2.Amulticast tree.

3.3 Problem Formulation

Our target is to develop an interference⁃aware routing algo⁃rithm.We need to construct themulticast tree according to the inter⁃and intra⁃network interferences.Then，we discuss the transmission and interferencemodel forour framework and for⁃mulate themulticast routing problem.The transmission and in⁃terferencemodel in this paper is similar to that in[7]，which in⁃cludesunicast routing over D2D networks.However，multicast⁃ing involves concurrent transmissions by different branches，and the interferencemodelneeds tobemodified accordingly.

We use P0to denote the transmission power of the cellular user，which is a constant.If D2D node i(i=1，2，...，U)is a transmitting node，ituses power Pi，which is regulated accord⁃ing to the interference statuses.All D2D node transmission power is characterized by a vector P={P1，P2，...，PU}.If a D2D node does not transmitdata during themulticast transmis⁃sion，its power is zero.The distance between the BSand cellu⁃lar user is denoted d0;the distance between node i and BS is denoted Di;we further useΔjto denote the distance between node j and cellular userwhile using di，jto denote the distance between node i and node j.While the D2D transmissions reuse the cellular user's uplink spectrum，the resulted signal⁃to⁃In⁃terference ratio(SIR)at the BScannotexceed a tolerable level，denotedρth.On the other hand，in order to guarantee the qual⁃ity of service(QoS)of the D2D communications，the data rate of each hop for D2Dmulticast，is required to be no less than a specified threshold Rth.

The transmissions at each hop for D2D multicast are per⁃formed in a time⁃division fashion.Beginning from the source node，the transmission for the x th hop occupies the x th time slot.When all destinations have received the data packet，thesource node begins tomulticast a new packet for all destina⁃tions.Future D2D networksmay work under the coordination and controlof the BS[1]，[2].So，weassume that the BScoordi⁃nate the entiremulticastprocessand share information such as the positionsand transmission poweramong D2D nodes.

Following the above principles，we now take the x th hop′s transmissions to establish themathematical framework.In this paper，we mainly consider the path loss for the transmission，where small⁃scale fading is assumed to be averaged by apply⁃ing diversity technologies.Correspondingly，the useful signal power received by BS，represented by G0，iswritten by

whereηrepresents the path⁃loss exponent andαis a con⁃stantdecided byantennagain and/orother factors.

Without loss of generality，we concentrate on the x th hop in themulticast transmission.Then，within the x th time slot，the SIR received at the BSstation，denoted by，isderived by

where Txis the transmitting node set of the x th hop.In order tosatisfy the interference constraint，i.e.，，weget

For any transmitting node in set Tx，the D2D transmission is affected by the cellular user′s signal and also the signals from other transmitting nodes in set Tx.The physical signals transmitted by multicast nodes can be different whereas the carried information is the same.Furthermore，it is hard to syn⁃chronize multicast nodes.Accordingly，other transmitting nodes cause interference as the cellular user does.Hence，the maximum achievable transmission rate from transmitting node i∈Txto its receivingnode j，denoted byℜi，j，can bewritten as

Following(4)，we can obtain themaximum transmission rate of the x th hop in the y th SL⁃path(i.e.，hop Hxy)，denoted byℜt，r，where t=Txy&r=Rxy，as

We design a routing algorithm that reduces the average hop count and lowers the outage probability.We define the hop count for a particular destination as the number of hops for it to receive amulticast packet from the source.Further，denot⁃ing the hop count for the v th destination by Lv，the average hop countoverallmulticastdestinations iscalculated by

Then we formulate a hop⁃countminimization problem as fol⁃lows:

whereκdenotes themulticast tree structure(see Definition 1) and P characterizes all D2D nodes'transmission power.The first constraint indicates the root node must be multicast source，the second constraint limits the interference caused to the cellular users'signal，the third constraint requires all hops formulticast can achieve the rate beyond Rth，the fourth con⁃straint implies that all leaf nodes be multicast receivers，and the lastconstraintsuggests thatanymulticast receiverneeds to beon some SL⁃path.

4 Angle-Based Interference-AwareRouting Algorithm for D2DMulticast Transm issions

The hop⁃countminimization problem formulated in section 3 involvesmixed integer⁃nonlinear programming，the optimal solution towhich isunrealistic to track.In this section，we pro⁃pose a heuristic algorithm，called angle⁃based interference⁃awaremulticast routing algorithm.Our algorithm has twomain parts.One is the receiving⁃node selection for each hop and the other is the power allocation for receiving nodes，which willbe respectively detailed in the following sections.Note thatwe as⁃sume that all location information for D2D nodes and cellular user is known to the BS.The BS will be responsible for the route selection and powerallocation for themulticastsession.

4.1 The Selection of Receiving Nodes for Each Hop

We take the following crucial factors into account for our routing design.First，the routes from the source need to ad⁃vance towards each specificmulticast destination.Second，the destinations are usually divided intomultiple groups lying on different branches of the multicast tree.Thus，each transmit⁃ting node often seeksmultiple receiving nodeswithin its trans⁃

5.2 Outage Probability

For a dynamic changingmulticast tree，it is difficult to ana⁃lyze the outage probability for routing because the routing nodes and transmission power are both uncertain.In this sec⁃tion，we analyze the outage probability for a hop.For a trans⁃mitting node i in Tx，the maximum transmission rate fromnode i to its receiving node j，denotedℜi，j，is given by(4).If we setℜi，j=Rth，we will get a closed curve.Nodes inside the region enclosed by the curve form the neighbor setof i.If there arenoactualnodes in the region，the routingbreaksoff.We de⁃note the regionσand the acreage ofσSσ，which is hard to calculate.Assume U D2D nodes are evenly distributed in a sector area with acreage S0.Then，the probability that there are no actual nodes in regionσ，i.e.，the probability for rout⁃ingoutage，is

From(11)we know that the outage probability will be lower if the regionσor the amount of D2D nodes is larger.The acreage ofσismainly determined by the transmission power and the strength of interference.Low transmission power or strong interference will reduce the acreage ofσ，which in⁃creases the outage probability.

5.3 The Im pactof Angle Threshold

The angle threshold is a very important for tree⁃building.It can determine whether two subsets should bemerged or not. Fig.4 isan example of differentmulticast routes resulted from different angle thresholdswith the same network topology.The threshold ranges from 0∘to 180∘.When the threshold is close to 0∘，themulticast tree expands intomultiple branches earli⁃er(Fig.4a).If the threshold is close to 180∘，the multiple branches will not be generated until the rest destinations are too far from each other(Fig.4b).

6 Simulation Evaluations

6.1 Two Modified Baseline Algorithm s

In this section，we describe the two baseline algorithmsused in the following simulation.One is the GMR routing algorithm in[17]，the other is the PBMrouting algorithm in[19].For the selection of receiving nodes for sxi，the original two algorithms both only consider the number of receiving nodes and further advance to destinations.However，the GMR algorithm involves amerging process of destinations like our proposed algorithm whereas the PBMalgorithm traverses all the possible subsets of Nito selectone as the setof receiving nodes.As the origi⁃nal two algorithms do not take into account the power con⁃straint in(3)，they cannotbe directly used in the D2D network. Therefore，wemodify the two algorithms tomake them suitable for our simulation scenario.The two modified algorithms are calledmodified⁃GMR(M⁃GMR)routing algorithm andmodified⁃PBM(M⁃PBM)routing algorithm.The only modification we make is to introduce the same allocation strategy described in section 4.2 to allocate power to eachselected by the two originalalgorithms.

6.2 Simulation Settings

In the simulation，we consider a sector area with a central angle equal to 2π/3 and radius500m，as illustrated in Fig.4. The transmission power P0used by cellular user is 23 dBm and the thresholdρthof SIR atBSis8 dB.The path⁃lossexpo⁃nentηis 3.The constantαis set such that the signal⁃to⁃noise ratio after 500 m transmissions is 0 dB.The positions of cellular user，D2D source node and D2D destination nodesare shown in Fig.4.Butother D2D nodes are generated randomly in the sector and the topology followsuniform distribution.The following simulation results are obtained by averaging over 1000 random ly generated topologies.

6.3 Simulation Results

◀Figure4. Routing exampleswith differentangle thresholds: (a)θth=45 and(b)θth=90 .

Fig.5a shows the outage probability versus the number U of D2D nodes.As described in[19]，the PBMalgorithm hasa pa⁃rameterλranging from 0 to 1，which also exists in the M⁃PBMalgorithm.We have thus run the same routing task 1000times.Only the best oneλ=0.8 is included in Fig.5.Here，we set Rth=2Mb/s andθth=110∘.As in Fig 5a，M⁃GMR and M⁃PBMhave much higher outage probability than our pro⁃posed algorithm because the routes generated by our proposed algorithm do notgonear the BSand regionwith strong interfer⁃ence whereas the others only consider providing advance to destinations.We can also observe from Fig.5a that the proba⁃bility of all three algorithms decreases as U increases.This phenomenon verifies the outage probability analysis in section 5.2.

Fig.5b shows the average hop⁃count result，where we set λ=0.8，Rth=2Mb/s andθth=110∘.The average hop⁃countof our proposed one is the leastamong three algorithms(Fig.5b). Meanwhile，the performance of M⁃GMR and M⁃PBMare get⁃ting closer when U becomes larger.This is because they both concentrate on finding nodes providing largestadvance to des⁃tinations.Thus，the two algorithms are inclined to choose the same routingpathwhen U is largeenough.

Fig.6a shows theoutage probability versus theangle thresh⁃oldθth，where we set Rth=2Mb/s and U=200.The outage probability ishighwhen the threshold issmallbecausewegen⁃erate toomany targetdestination subsetsand corresponding re⁃ceiving nodes.As in(3)，the transmission power and transmis⁃sion range of each node will be quite limited，which will in⁃crease the outage probability.The outage probability decreases whenθthincreases and trends towards a stable value.This is because wemerge all destinations into one target subset and the multicast tree has hardly any expanded branches.Thus，the outage probability ismainly decided by the number of D2D nodes in the network.

Fig.6b shows the average hop⁃count versus the angle thresholdθth.The small threshold leads tomore hops because we generate toomany next⁃hop nodes.Their transmission rang⁃es are quite limited，as discussed in the last paragraph.Thus，the average hop⁃count is high.The hop⁃countdecreases asθthincreases.Whenθthis larger than 85∘，the hop⁃count increas⁃esagain and trends to a stable value.Therefore，85∘is the best threshold for the specific positions of cellular user，D2D source node and D2D destination nodesgiven in Fig.4.

◀Figure5. Performance comparisons for three routing algorithms:(a)Outage probability versus the number U ofD2D nodes, and(b)Averagehop⁃count versus thenumber U of D2D nodes.

◀Figure6. Performanceevaluations for our proposed algorithm: (a)Outageprobability versusangle threshold,and (b)Averagehop⁃count versusangle threshold.

7 Conclusion

In this paper，we proposed an angle⁃based interference⁃aware routing algorithm for multicast over wireless D2D net⁃works.Our proposed algorithm aims to lower the outage proba⁃bility and minimize average hop⁃count.By utilizing the DRMprinciple and the angle⁃threshold basedmerging principle，the packets can effectively progress towards destinations.Our pro⁃posed algorithm has low computational complexity.Simulation results show thatour proposed algorithm outperforms the given baseline schemes in terms of outage probability and average hop count.

[1]K.Doppler，M.Rinne，C.Wijting，C.B.Ribeiro，and K.Hugl，“Device⁃to⁃device communication as an underlay to LTE⁃advanced networks，”IEEE Commun. Mag.，vol.47，no.12，pp.42-49，Dec.2009.doi:10.1109/ MCOM.2009.5350367.

[2]G.Fodor，E.Dahlman，G.Mildh，S.Parkvall，N.Reider，G.Miklos，and Z. Turanyi，“Design aspectof network assisted device⁃to⁃device communications，”IEEE Commun.Mag.,vol.50，no.3，pp.170-177，Mar.2012.doi:10.1109/ MCOM.2012.6163598.

[3]H.Min，J.Lee，S.Park，and D.Hong，“Capacity enhancementusing an interfer⁃ence limited area for Device⁃to⁃Device uplink underlaying cellular networks，”IEEE Trans.Wireless.Commun.，vol.10，no.12，pp.3995-4000，Dec.2011. doi:10.1109/TWC.2011.100611.101684.

[4]H.Wang and X.Chu，“Distance⁃constrained resource⁃sharing criteria for device⁃to⁃device communications underlaying cellular networks，”Electron.Lett.，vol. 48，no.9，pp.528-530，Apr.2012.doi:10.1049/el.2012.0451.

[5]H.Min，W.Seo，J.Lee，S.Park，and D.Hong，“Reliability improvementusing re⁃ceivemode selection in the device⁃to⁃device uplink period underlaying cellular networks，”IEEE Trans.Wireless.Commun.，vol.10，no.2，pp.413-418，Feb. 2011.doi:10.1109/TWC.2011.122010.100963.

[6]Y.Cheng，Y.Gu，and X.Lin，“Combined power controland link selection in de⁃vice⁃to⁃device enabled cellular systems，”IET Commun.，vol.7，no.12，pp. 1221-1230，Aug.2013.doi:10.1049/iet⁃com.2012.0769.

[7]P.Ren，Q.Du，L.Sun，“Interference⁃aware routing for hop⁃countminimization in wireless D2D networks，”in IEEE/CIC ICCC 2013 Workshop Internet of Things，Xi’an，China，2013，pp.65-70.doi:10.1016/j.adhoc.2008.02.004.

[8]J.Wang，D.Zhu，C.Zhao，J.C.F.Li，and M.Lei，“Resource sharingofunderlay⁃ing device⁃to⁃device and uplink cellular communications，”IEEE Commun. Lett.，vol.17，no.6，pp.1148-1151，Jun.2013.doi:10.1109/ LCOMM.2013.042313.130239.

[9]C.⁃H.Yu，K.Doppler，C.B.Ribeiro，and O.Tirkkonen，“Resource sharing opti⁃mization for device⁃to⁃device communication underlaying cellular networks，”IEEE Trans.Wireless.Commun.，vol.10，no.8，pp.2752-2763，Aug.2011.doi: 10.1109/TWC.2011.060811.102120.

[10]P.Pahlevani，M.Hundebøll，M.V.Pedersen，D.Lucani，H.Charaf，F.H.P. Fitzek，H.Bagheri，and M.Katz，“Novel concepts for device⁃to⁃device commu⁃nication using network coding，”IEEECommun.Mag.，vol.52，no.4，pp.32-39，Apr.2014.doi:10.1109/MCOM.2014.6807944.

[11]J.Luo，D.Ye，L.Xue，and M.Fan，“A survey ofmulticast routing protocols for mobile Ad⁃Hoc networks，”IEEE Commun.Surveys.Tutorials，vol.11，no.1，pp.78-91，Mar.2009.

[12]D.D.Le，M.Molnár，and J.Palaysi，“Multicast routing inWDMnetworkswith⁃out splitters，”IEEE Commun.Mag.，vol.52，no.7，pp.158-167，Jul.2014. doi:10.1109/MCOM.2014.6852098.

[13]C.Gao，Y.Shi，Y.T.Hou，H.D.Sherali，and H.Zhou，“Multicast communica⁃tions inmulti⁃hop cognitive radio networks，”IEEE J.Sel.AreasCommun.，vol. 29，no.4，pp.784-793，Apr.2011.doi:10.1109/JSAC.2011.110410.

[14]H.M.Almasaeid，T.H.Jawadwala，and A.E.Kamal，“On⁃demand multicast routing in cognitive radiomesh networks，”in IEEEGLOBECOM2010，Miami，USA，pp.1-5.doi:10.1109/GLOCOM.2010.5683665.

[15]Z.Chen，D.Jiang，Z.Xu，Y.Han，H.Xu，and P.Zhang，“Amulticast routing al⁃gorithm in cognitive ad hoc networks，”in ICCP 2010，Lijiang，China，pp.284-288.

[16]S.Wu and K.S.Candan，“GMP:distributed geographic multicast routing in wireless sensor networks，”in ICDCS 2006，Lisboa，Portugal，pp.1-9.doi: 10.1109/ICDCS.2006.44.

[17]J.A.Sanchez，P.M.Ruiz，I.Stojmenovic，“GMR:geographicmulticast routing forwireless sensor networks，”in SECON 2006，Reston，USA，pp.20-29.doi: 10.1109/SAHCN.2006.288405.

[18]J.Cha，J.Jeon，J.Kim，and Y.Kwon，“Location⁃based multicast routing algo⁃rithms forwireless sensor networks in presence of interferences，”in ICCS，Sin⁃gapore，2010，pp.41-45.doi:10.1109/ICCS.2010.5686104.

[19]M.Mauve，H.Füβler，J.Widmer，T.Lang，“Position⁃based multicast routing for mobile ad⁃hoc networks，”ACMSIGMOBILE Mobile Comput.Commun. Rev.，vol.7，no.3，pp.53-55，Jul.2003.doi:10.1145/961268.961288.

Biographies phies

Qian Xu(xq1216@stu.xjtu.edu.cn)received her BSdegree in information engineer⁃ing from Xi’an Jiaotong University，China，in 2014.She is currently working to⁃wards the PhD degree in communication and information system at the same univer⁃sity.Her research interests include D2D networksand wireless communications.

PinyiRen(pyren@mail.xjtu.edu.cn)received his BS，MSand PhD degrees from Xi’an Jiaotong University，China.He is currently a prefessor and the departmenthead of Information and Communications Engineering Department，Xi’an Jiaotong Uni⁃versity.His current research interests include cognitive radio networks，MIMO sys⁃tems，game theory in wireless communications，wireless relay，routing，and signal detection.He has publishedmore than 80 technical papers in international Journals and conferences.He received the Best Letter Award of IEICECommunications Soci⁃ety in 2010.He hasmore than 10 authorized Chinese Patents.Prof.Ren serves as an editor for the Journal of Xi’an Jiaotong University，and served as the leading guesteditors for the special issue of Mobile Networksand Applications on“Distribut⁃ed WirelessNetworksand Services”and thatof Journal ofElectronics on“Cognitive Radio”.

Qinghe Du(duqinghe@mail.xjtu.edu.cn)received his BSand MSdegrees both from Xi’an Jiaotong University，China，and his PhD degree from Texas A&MUniversity，USA.He is currently an assistant professor of Information and Communications En⁃gineering Department，Xi’an Jiaotong University，China.His research interests in⁃clude mobile wireless communications and networking with emphasis on mobile multicast，statistical QoS provisioning，and cognitive radio networks.He has pub⁃lished more than 30 technical papers.He received the Best Paper Award in IEEE GLOBECOM2007.He serves as an associate editor of IEEE Communications Let⁃ters.

GangW u(wu.gang26@zte.com.cn)received his PhD degree from SoutheastUniver⁃sity，China in 2002.He is currently a R&D expertat ZTECorporation.His R&D in⁃terests include wireless communication system and terminal chipsetwith emphasis on algorithm，and system design and standardization.He has published more than 20 technical papers，40 international patents and 10 3GPP standardization propos⁃als.He is leading a research task of National Science and Technology Major Project of China.

Qiang Li(li.qiang8@zte.com.cn)received his PhD degree in communications&in⁃formation systems from Southeast University，China.He is currently the head of Al⁃gorithm Design Department of ZTE Corporation，and responsible for the research and design of telecommunication baseband algorithms.

Li Sun(lisun@mail.xjtu.edu.cn)received his BS and PhD degrees in Information Engineering from Xi’an Jiaotong University，China，in 2006 and 2011.He is cur⁃rently an assistant professor at the School of Electronic and Information Engineer⁃ing，Xi’an Jiaotong University，China.His research interests include cooperative re⁃laying networksand wireless communications.

t

2014⁃08⁃25

10.3969/j.issn.1673-5188.2014.04.005

http://www.cnki.net/kcm s/detail/34.1294.TN.20141212.0940.002.htm l,pub lished online Decem ber 12,2014

This wo rk is suppo rted by National Natu ralScience Foundation o f China under Gran t No.61102078,ZTE Industry⁃Academ ic⁃Research Cooperation Funds,and the Fundam en tal Research Funds fo r the Cen tralUniversities.