A Novel Routing Algorithm for Power Line Communication over a Low-voltage Distribution Network in a Smart Grid

Transcription

1 Energies 0, 6, 4-48; doi:0.90/en604 Aricle OPEN ACCESS energies ISSN A Novel Rouing Algorihm for Power Line Communicaion over a Low-volage Disribuion Nework in a Smar Grid Liang Zhang, Xiaosheng Liu *, Yan Zhou and Dianguo Xu Harbin Insiue of Technology, Harbin 5000, China; s: xiaozhanghi@6.com (L.Z.); zhouyan@hi.edu.cn (Y.Z.); xudiang@hi.edu.cn (D.X.) * Auhor o whom correspondence should be addressed; liuxsh004@6.com; Tel./Fax: Received: 9 November 0; in revised form: 8 January 0 / Acceped: February 0 / Published: 5 March 0 Absrac: A novel arificial cobweb rouing algorihm (ACRA) for rouing he ree-ype physical opology of a low-volage disribuion nework in a smar grid is proposed and analyzed in his paper. The esablishmen, mainenance and reconsrucion of he roue are presened. The arificial cobweb rouing algorihm is shown o have broad general applicabiliy for power line communicaion. To provide a heoreical foundaion for furher research, he communicaion delay of he nework is calculaed accuraely. Simulaion analysis of he communicaion delay and hroughpus, which were based on Opne4.5, demonsrae he accuracy of he heoreical calculaion. For he performance evaluaion of ACRA, a es-bed ha includes PLC nodes wih he ACRA is se up in a noisy environmen. Experimenal resuls show he feasibiliy of he ACRA algorihm. These indicae ha ACRA is effecive for guaraneeing Qualiy of Service (QoS) and reliabiliy in power line communicaion. Keywords: smar grid; power-line communicaion; arificial cobweb rouing algorihm; reliabiliy; qualiy of service. Inroducion Recenly, he Sae Grid Cooperaion Company of China, which is a sae-owned company, decided o consruc new grids based on Smar Grid echnology across China. Power-line communicaion (PLC) over he low-volage (LV) disribuion neworks has become one of he poenial echnologies o

2 Energies 0, 6 4 communicae informaion beween end users and power providers, and hus, recen research in China has focused on PLC []. A presen, research on PLC focuses primarily on he aspec such as inpu impedance and aenuaion characerisics of he channel [], channel noise characerisics [,4], channel models [5,6], impedance maching and coupling circui design [7], power line noise suppression echnology [8,9], he applicaion of orhogonal frequency division muliplexing (OFDM) in PLC, frequency hopping modulaion and demodulaion echniques [0]. An effecive mehod o improve he reliabiliy of PLC sysems from nework level is hrough he esablishmen of nework rouing or relay echniques [,]. An arificial cobweb rouing algorihm for rouing he ree-ype physical opology of a low-volage disribuion nework is analysed in his paper based on he arificial cobweb rouing of he bus-ype daa communicaions layer of a low-volage disribuion nework proposed in []. The arificial cobweb rouing algorihm is furher improved. To provide a heoreical foundaion for furher research, he nework parameers afer neworking are accuraely calculaed. The simulaion resuls demonsrae he accuracy of he heoreical calculaion. The resuls indicae ha he arificial cobweb rouing algorihm is effecive for guaraneeing QoS and reliabiliy in power line communicaion.. Arificial Cobweb Rouing Algorihm.. Topology of Low-Volage PLC For a hree phase power disribuion grid, he signal aenuaion beween phases on he secondary side of a ransformer is very large. Wihou phase coupling, each phase can be regarded as parallel and relaively independen; hus, he opology of any one phase can be he focus of research. On he basis of he physical opology of a low-volage disribuion nework proposed in reference [4], his paper esablishes a single-phase power disribuion nework, as shown in Figure. The base saion (BS) is responsible for he collecion and coordinaion of he daa from each erminal node in he nework; he BS is conneced wih a wide area nework (WAN) for exernal informaion exchange. In pracice, he signal is aenuaed as a resul of he ransmission disance [5]; herefore, only he erminal nodes wih a shor physical disance from he BS are able o ensure reliable communicaion. Over long ransmission disances, radiional polling mehods canno guaranee he direc communicaion beween he BS and he nodes far away from he BS, so he success rae of communicaion may be very low [4]. To solve his problem, a novel mehod called he arificial cobweb rouing algorihm is proposed. In his mehod, a communicaion roue is firs esablished beween he BS and some of he nodes. The nodes are hen regarded as relays o expand he communicaion disance, and gradually, all he nodes are conneced o he communicaion sysem. Finally, he roue beween he BS and all he nodes is compleed. Consequenly, i is necessary o selec he relay nodes and esablish, mainain and reconsruc he roue.

3 Energies 0, 6 4 Figure. Tree-ype nework. BS Arificial Cobweb Rouing Algorihm A novel rouing algorihm based on he single-layer arificial cobweb srucure presened in reference [] is proposed. In a single-layer arificial cobweb srucure as shown in Figure, he cenral nodes (such as node I in Figure ) are responsible for coordinaing and dealing wih he informaion from he peripheral nodes (such as node,,,, h in Figure ) wihin a subne. In addiion, cenral nodes are relays for communicaion beween he cenral nodes and BS of oher associaed subnes. The communicaion roue of a BS o any one erminal node is exclusive in a nework wih an arificial cobweb srucure. Figure. Single-layer arificial cobweb srucure. I h 4 5 Assuming ha he physical link is conneced as shown in Figure, we sipulae ha:. All nodes in he neworking ess are responsible for recording he physical signal srengh β of he received informaion;. Any node wihin he nework can communicae wih a leas one oher node;. Each node aemps o find he possible number of nodes in a maximum physical range; 4. The logical ID of he BS is 0, and he nodes which have been assigned logical ID is no longer involved in he new logical ID allocaion ess; 5. Afer compleing he neworking, he cenral node will no become a bad node (The nodes ha canno communicae wih oher nodes as a resul of changes o he channel condiion).

4 Energies 0, Arificial Cobweb Rouing Iniializaion Algorihm The flow char of arificial cobweb rouing algorihm is shown in Figure. In wha follows, we illusrae he arificial cobweb rouing iniializaion algorihm combined wih Figure and Figure : Figure. Flow char of algorihm. Begin Iniializaion BS/cenral node send neworking broadcas Nodes received he broadcas send response o BS/cenral node in urn Assigned ID by he BS/cenral node Send response o BS/cenral node again No BS/cenral node selec a node in hese nodes as new cenral node The new cenral node send broadcas and received by he oher nodes which have go ID Nodes receive he broadcas from new cenral node, record he ID of new cenral node new_ cenral_ node BS / cenral_ node New cenral node generae rouing able and send i o BS/cenral node New cenral node send neworking broadcas again Are here nodes wihou ID received he broadcas? Yes No Yes These nodes Send response o new cenral node Roue iniializaion is compleed End. When a neworking broadcas is sen by he BS, in order o explain he ess of algorihm convenienly, we assume ha i can only be inerceped by nodes o 9, which have a close physical disance o he BS. These nodes send responses o he BS in urn and are assigned logical IDs ( o 9) by he BS. A broadcas from he BS is received by nodes o 9, which indicaes ha he link in he physical range of hese nodes is in good condiion and ha he 9 nodes can direcly or indirecly communicae wih each oher. BS selecs a node h from hese 9 nodes as he cenral node.. Node h sends a broadcas. The res of he eigh nodes receive he broadcas, record he logical ID of h (nodes wihou a logical ID do no respond o he broadcas), and send a response o h. Cenral node h records he logical IDs of peripheral nodes o generae a rouing able. The rouing able is sen o he BS so ha he arificial cobweb opology nework of subne (single-layer arificial cobweb srucure) is compleed and records ha he logical layer where he nine nodes are locaed, is layer.. Node h sends a neworking broadcas, assuming ha he broadcas is received by nodes 0~4, wih he logical layer conaining hese nodes recorded as layer. Repeaing sep ), node h assigns a logical ID and selecs cenral node h, assuming ha h is any one of nodes 0~6. h sends a broadcas and repea sep (). Nodes 7~4, which are locaed in differen branches, do no receive he broadcas from h, or if hey do receive a broadcas, he broadcas is weak in

5 Energies 0, 6 45 inensiy as a resul of signal aenuaion. In his case, he composie index of signal srengh β is inroduced o make a furher judgmen. If β h < β h, hen he composie index of signal srengh from h recorded by nodes 0~6 is less han ha from h, sep is repeaed, h is seleced as he cenral node o form a new nework. If β h > β h or a broadcas from h is no received, hen nodes 7~4 send a message o h and repea sep (), and h selecs a new cenral node h from hese nodes o form a new cobweb nework. This paper assumes ha a broadcas from h is no received by nodes 7~4, which are locaed in anoher branch, and h is seleced as he cenral node o form a new cobweb nework. I is sipulaed ha he logical layer of he new cobweb nework is layer. h and h send he rouing able o he BS by selecing h as a relay. β = s η l, where s is he characerisaion value of he physical signal srengh, l is he layer number of he subne o which he informaion sources belong, and η is he successful probabiliy of communicaion beween differen nodes. η can be obained from pracical saisical values. A higher value for η indicaes higher reliabiliy of communicaion beween layers. Conversely, a lower value for η indicaes lower reliabiliy. 4. h and h send a broadcas again, assuming ha nodes 5~6 receive he broadcas from h, repea sep, and selec cenral nodes h 4 and h 5 o form new cobweb neworks in differen branches. When here are no nodes wihou a logical ID response o broadcas from h and an empy response is obained, hen sop neworking. 5. h 4 and h 5 send a broadcas and repea sep (). When here are no longer nodes wihou logical ID response o broadcas from h 4 and h 5 and an empy response is obained, and hen sop neworking. Sipulae ha layer is he spider web logical layer ha h 4 and h 5 locae. h 4 and h 5 send he roue able o he BS by selecing h and h as relays. Finally, he iniializaion is compleed. If he cenral nodes h 4 and h 5 of he hird layer sill receive nodes wihou a logical ID, hen hey repea he neworking ess unil all nodes have a logical ID and are conneced o he nework. The communicaion roue of all nodes wihin he nework will hen be esablished.... Example of he Rouing Algorihm The single-phase power disribuion nework in Figure consiss of one BS and 6 erminal nodes. Afer he arificial cobweb iniializaion algorihm, he MAC layer of he nework reflecs he nework srucure shown in Figure 4. One radio broadcas of he BS is inerceped by nine nodes, and hese nine nodes will hen respond o he BS one by one o receive a logical ID. The BS selecs he node, we assume he logical ID 5, o be he cenral node of layer, and hen, hese nodes compose a cobweb in he MAC layer based on he seps,. The node wih he logical ID 5 coninues o broadcas and is inerceped by 5 nodes from differen physical branches, which are hen assigned logical IDs 0~4. The node wih he logical ID is seleced as he cenral node for neworking. In case he broadcas of his cenral node is no inerceped by a node wih a logical ID assigned in anoher branch, his node will send a message o node 5. Node 5 will hen reselec node 9 as he cenral node o complee he nework of layer. According o he acual environmen and redundan calculaion of ime delay, we sipulae ha node 5 wai 0 s o collec all he replies before esablishing cenral nodes for layer.

6 Energies 0, 6 46 Cenral nodes and 9 search wih he same algorihm unil he res of he nodes obain logical IDs o complee he nework in layer. Figure 4. Resul of neworking. BS subne 5 9 Layer subne subne Layer Layer subne4 subne5 Afer compleing he cobweb roue iniializaion, each cenral node records he logical ID and layer of he nodes in he same subne. The peripheral nodes record he logical IDs and layers of heir own nodes and he cenral nodes. The complexiy and lengh of he rouing able is deermined by he number of he nodes and he layers in he nework.. Rouing Mainenance and Reconsrucion Algorihm.. Rouing Mainenance Principle Because of he openness and ime variabiliy of he low-volage power line nework, bad nodes will appear during he operaion of he nework. Two principles o mainain he roue are proposed o ensure he success rae of he daa collecion for he BS: when bad nodes caused by rouing inerrup he operaion of he compleed nework, a ess o reconsruc he local rouing of he bad nodes begins (in Secion.). during he idle periods of he nework, he BS iniiaes he rouing deecion insrucions. If a bad node is found, principle () is implemened o esablish new communicaion roues. In he secondary side of a 00 m 00 m ree-ype grid, we se 9 end-user nodes and a base saion node (Node ) in Figure 5. The resuls for he arificial cobweb rouing algorihm (ACRA) are shown in Figure 6. Each node in a grid wih ree opology only has a single poin-o-poin link shown in Figure 5. If one communicaion link (such as links -) is inerruped, unless he enire nework is reconsruced, he inerruped communicaion link canno be resored. In he ACRA nework shown in Figure 6, he nodes in he same layer (such as nodes, 8,, 6) can communicae wih oher nodes. A peripheral node may coninue o communicae wih cenral node owing o he addiional angenial

7 Disance(m) Disance(m) Energies 0, 6 47 connecions when one communicaion link is inerruped. For example, when he links 8 are inerruped, peripheral node 8 can communicae wih cenral node by choosing peripheral node as a relay. Afer neworking, he BS can communicae wih any node in he neworking via he cenral node, which acs as a relay node. An example is shown in Table. I is obvious ha he communicaion qualiy is more effecive using he ACRA han using ree opology. The ACRA can effecively reduce he nework reconfiguraion frequency. 00 Figure 5. Tree opology model Disance(m) Figure 6. Resuls of ACRA Disance(m).. Rouing Reconsrucion Algorihm Table. Example of rouing. Source node Desinaion node Relay nodes 5 7, 0 9 7, 4, 5 As shown in Figure 4, we assume ha he cenral node in logical layer has no received any informaion sen by node, afer a daa essing delay, confirming ha node is a bad node. Node hen iniiaes he following ess o reconsruc he local rouing:

8 Energies 0, Cenral node sends a reconsrucion broadcas o he nodes in is same subne.. Nodes ha inercep he insrucions for rouing reconsrucion forward he insrucions in urns and add heir own logical ID ino he daa packe.. Node records he physical srengh of he reconsrucion broadcas signals; calculaes and selecs he node wih max β, he composie index of he signal srengh, as a relay node (node in his paper); and sends a response o node. Node esablishes a communicaion roue o cenral node wih node as relay. 4. Node sends informaion o node, and node updaes he rouing able. 5. Node sends he informaion layer by layer up o he BS of he corresponding node o updae he rouing able. Then, he reconsrucion ess for rouing a bad node is compleed. When bad nodes appear in a nework, i is no necessary for all nodes o reconsruc heir rouing. Only he nodes in he same subne as he bad nodes are involved, which dramaically shorens he roue reconsrucion and mainenance ime and improves he efficiency of he sysem... Algorihm Timing PLC only allows one node o send daa a any given ime, so i is necessary o presen a reasonable iming sraegy. Oherwise, a channel conflic will seriously affec he performance of he sysem. Using he daa ransmission iming of he nodes in a subne, he arificial cobweb rouing algorihm (ACRA) is compared wih he clusered simple polling (CSP, developed from he general polling mehod used o solve he silen node problem in he PLC-access nework) algorihm menioned in reference [] hrough simulaion. The inerval ime of communicaion beween wo peripheral nodes T WT in he ACRA is defined as (). T P represens he manage essing delay of one peripheral node; T T is he signal ransmission ime among nodes; T R indicaes he communicaion redundancy ime. Generally, he signal ransmission ime T T in a power line is fas enough o be ignored: T WT Tp TT TR () Based on he proocol and he experimenal resuls [6,7], T WT is assumed o be 50 ms in his paper. Finally, he communicaion effeciveness of a cobweb srucure is confirmed using daa collision and he channel uilisaion percenage. Figure 7a shows he iming of he ACRA, in which he horizonal axis is ime, he single-layer cobweb consiss of n nodes, node n is he cenral node, he res of he nodes are peripheral nodes, and every node sends a daa message in urn, according o he smalles o larges logical ID, o cenral node of is subne wih an inerval ime T WT. Afer receiving messages from all nodes, he cenral node hen sends he messages o he cenral node or BS. This ess avoids he channel daa collision caused by he occupaion of a channel by several nodes a he same ime. Figure 7b shows he iming of CSP using five nodes. A node of medium disance will communicae wih he BS hrough a closer node as a relay. Wihou consan relay nodes and sric iming, he five nodes will find he close node by sending random messages, resuling in a large amoun of daa collision in he channel. The corresponding simulaion resul is described in deail in he fourh par of his paper.

9 Energies 0, 6 49 Figure 7. Timing of he algorihms. (a) Timing of he ACRA; (b) Timing of CSP. node WT node node node WT WT node node Collision Collision Collision Collision node m (m-)wt node4 Collision node n PT (ms) node5 (ms) (a) (b) 4. Delay Analysis 4.. Delay Analysis wihou Bad Nodes Communicaion delay is an imporan measure of QoS, so i is necessary o calculae he communicaion delay of he arificial cobweb nework accuraely. In he ree opology, here is only one link conneciviy beween any wo nodes, node communicaion failure due o channel noise and oher uncerainies, which lead o he ime of recovery communicaion canno be accuraely calculaed, he delay ime may be emporary or permanen. So he ime of ree opology is no calculaed in his paper. The ransmission ime of he packe is available from Equaion (): p / Baud () where p is he packe size and Baud is he communicaion rae. This paper ses he daa essing delay = 0.5 s. Afer neworking, he only facor ha affecs he communicaion delay is he subne ha he nodes come from. The communicaion delay is he same for differen subnes ha selec he same cenral node as he relay node. Therefore, we selec subne, subne, and subne 4 from layer, layer, and layer, respecively, in Figure 4 o calculae he delay. Firs, we calculae he delay of peripheral node s, and all he nodes in subnes wihou bad nodes communicae wih he BS. The seps of he ransmission of daa from a peripheral node in subne o he BS shown in Figure 8 are as follows: (a) The BS sends a daa acquisiion command o he cenral node h of subne. (b) A peripheral node in subne sends is daa o h. (c) Afer he daa essing ime, node h sends he daa of a peripheral node o he BS. The delay d in he ransmission of daa from a peripheral node in subne o he BS is shown in (): d () The seps of he ransmission of daa from a peripheral node in subne o he BS are as follows: (a) The cenral node h sends a daa acquisiion command o he cenral node h of subne, and a peripheral node in subne sends is daa o h. (b) Afer he daa essing ime, node h sends he daa of a peripheral node o h. (c) Afer he daa essing ime, node h sends he daa o he BS.

10 Energies 0, 6 40 Figure 8. Process of ransmission delay wihou bad node. BS Subne Subne Subne 4 h i 5 h h The delay d in he ransmission of daa from a peripheral node in subne o he BS is shown in Equaion (4): d 4 (4) The ess is similar for he nodes in subne and subne. The delay in he ransmission of daa from a peripheral node in subne 4 o BS d is shown in Equaion (5): d 5 (5) Similarly, he delay in he ransmission of daa from a peripheral node in subne m of layer n o BS dn is shown in Equaion (6): n n n dn (6) We assume ha subne conains p nodes: he seps of he ransmission of daa from all he nodes in subne he BS are as follows: (a) BS sends he daa acquisiion command o he cenral node h of subne. (b) The p peripheral nodes send daa o h in urns. (c) Afer he daa essing ime, node h sends all he daa of p nodes (include he daa of node h iself) o he BS. According o he calculaive seps, he delay in he ransmission of daa from all he nodes in subne o BS T d is shown in Equaion (7): T p p p d (7) We assume ha subne conains p nodes and ha subne 4 conains p nodes. Similar o he calculaion essing of T d, he delay in he ransmission of daa from all he nodes in subne and subne 4 o BS T d and T d are shown in Equaions (8) and (9), respecively: p p p p Td (8) p p p 4p Td (9) Similarly, he delay in he ransmission of daa from all he nodes in subne m of layer n, which conains p n nodes, o BS T dn is shown in Equaion (0): T n p n n d n n (0) In summary, assuming ha each layer conains only one subne, afer neworking, he delay in he ransmission of daa from all he nodes o BS T dall is shown in Equaion ():

11 Energies 0, 6 4 T dall n i T p 4 p p p 4 p n p m n di p n p n n n m n n () 4.. Delay Analysis wih Bad Nodes According o he rouing reconsrucion algorihm proposed in his paper, he delay in he ransmission of daa from one bad node in subne, subne and subne 4 o BS d, d, and d is shown in Equaions (), (), and (4), respecively. The ess of ransmission delay is shown in Figure 9. p p p - p d () p p p d p () p p p p d (4) Figure 9. Process of ransmission delay wih bad node. BS j Layer.Subne Layer.Subne Bad node Layer.Subne4 4 h i 5 k h h l Similarly, he delay in he ransmission of daa from one bad node in subne m of layer n o BS dn is shown in Equaion (5): p n - p n n dn n n (5) According o he rerouing algorihm, here is one bad node each in subne, subne and subne 4, respecively. The delay in he ransmission of daa from all he nodes o BS T d, T d, and T d is described by Equaions (6), (7) and (8), respecively. p d Td T 4 (6) p d Td T 5 (7) 4 p d Td T 6 (8) Similarly, if subne m in layer n conains w- good nodes and one bad node, hen he delay in he ransmission of daa from all he nodes in subne m o BS T dn is described in Equaion (9): n p n n Tdn n (9) In summary, when each subne conains one bad node, he delay in he ransmission of daa from all he nodes o he BS based on he rerouing algorihm T dall is described in Equaion (0):

12 Delay(s) Energies 0, 6 4 T dall n i T di p p 4 p n p n n p p 4 p n p n m n n (0) Figure 0 shows he calculaed resuls based on he Equaions proposed above, where each subne conains seven nodes and he subne layer n is from o 0. I shows ha when n = 0 and each subne conains bad node, T dall is abou 68 s. Wih an auomaic meer reading sysem, which is less demanding on he iming, he delay can fully guaranee he QoS of sysem. Though his mehod coss some ime, i guaranees he daa collecion success rae of he BS, and has he desirable pracicabiliy. These resuls demonsrae ha even hough he mehod is ime consuming, he ACRA is effecive for guaraneeing QoS in PLC. n n m Figure 0. Calculaed resuls of delay T dall T dall dn dn n 5. Simulaion and Experimen 5.. Simulaion Environmen and Parameer To simulae he disribuion environmen of an acual low-volage disribuion nework, we se erminal nodes and one BS node wihin 50 meers range radius and used a PC as he simulaion plaform, wih Opne4.5 as he compilaion and simulaion environmen. We assume ha all nodes form hree arificial cobwebs and ha his srucure remains unchanged hroughou he simulaion period. Figure shows he opology srucure afer neworking, in which subne_0 represens he BS node, subne 0, subne 0 and subne 0 are he cenral nodes of he subnes, and he oher nodes are erminal nodes. Based on he Konnex sandard [8], we se he channel ransmission rae o.4 kbps and he packe size o 4 bis.

13 Throughpu(bis/s) Delay(s) Energies 0, 6 4 Figure. Simulaion model. subne subne subne subne_0 subne subne subne subne subne subne subne subne 0 subne 4 subne subne 0 subne 4 subne subne 0 subne 4 subne 6 subne 5 subne 6 subne 5 subne 6 subne Simulaion Resuls and Analysis Wih he 4 bis/s generaing rae of each node, he calculaed hroughpu of link subne 0-subne 0 is 68 bis/s. The hroughpu of link subne 0-subne 0, including he daa of subne and subne, is 6 bis/s. The hroughpu of link subne 0-subne_0, which includes he daa of all he nodes, is 504 bis/s. The hroughpu simulaion resuls of links subne 0-subne 0, subne 0-subne 0 and subne 0-subne_0 are shown in Figure a. The figure shows ha he hroughpu simulaion resuls of every link are idenical o he heoreical calculaions. According o Equaion (), he calculaed ransmission delay of each link is 0.07 s, 0.4 s, and 0. s. The simulaion resuls of he daa ransmission delay beween each link are shown in Figure b. The delay simulaion resuls are also idenical o he heoreical calculaion. The simulaion resuls indicae ha wih a hroughpu and ransmission delay, he neworking srucure can guaranee QoS in power line communicaion. Figure. Simulaion resuls of links beween cenre nodes. (a) Throughpu beween cenral nodes; (b) Delay beween cenral nodes subne 0-subne subne 0-subne subne 0-subne 0 0. subne 0-subne subne 0-subne subne 0-subne Simulaion Time(s) (a) Simulaion Time(s) (b) According o he simulaion model shown in Figure, he parameers of Equaions (6), (), (5), and (0) are n =, j = 7, k = 7 and l = 7, and he corresponding delays are d =.55 s, d = 5.5 s, T dall =.6 s, and T dall = 4.4 s. The simulaion resul of he delay in he ransmission of daa from node subne o he BS is shown in Figure a. I shows ha he delay is beween.55 s and.56 s; his resul corresponds wih he heoreical calculaion. In he simulaion model, we selec ha subne canno communicae wih subne 0, which is he cenral node of subne. The rerouing algorihm is used, and subne is he relay. The simulaion resul of he delay in he ransmission of

14 Delay(s) Delay(s) Delay(s) Delay(s) Energies 0, 6 44 daa from node subne o he BS is shown in Figure b. The delay is beween 5.4 s and 5.6 s; his resul corresponds wih he heoreical calculaions. The simulaion resul of he delay in he ransmission of daa from all he nodes o he BS is shown in Figure c. The delay is beween.6 s and.6 s; his resul is consisen wih he heoreical calculaions. The simulaion resul of he delay in he ransmission of daa from all he nodes o he BS when subne, subne and subne have bad nodes and hey selec adjacen nodes as a relay is shown in Figure d. The resul is consisen wih he heoreical calculaions. The simulaion resuls shown in Figure demonsrae he accuracy of he delay heoreical calculaion previously menioned. The simulaion also shows ha alhough he rerouing algorihm is ime consuming, i increases he daa collecion success rae of he BS, guaranees QoS in power line communicaion, improves he reliabiliy of he sysem and is pracical. Figure. Simulaion resuls of delay. (a) Simulaion resuls of d ; (b) Simulaion resuls of d; (c) Simulaion resuls of T dall ; (d) Simulaion resuls of T dall Simulaion Time(s) (a) Simulaion Time(s) (c) Simulaion Time(s) (b) Simulaion Time(s) (d) Figure 4a shows a comparison of he simulaion resuls relaed o channel conflic beween ACRA and CSP. Verical lines represen he occurrence of channel conflic a a specific ime. The figure shows ha wihin he simulaion ime, conflic caused by ACRA (blue line) is far less han ha caused by CSP (green line). ACRA divides he whole large nework ino several small neworks, in which he nodes of each nework send messages o a cenral node wihin a cerain ime. This ess reduces he nodes in he channel a any one momen, which considerably reduces he conflic and improves he channel uilisaion, as shown in Figure 4b.

15 Channel Conflic Uilizaion Percen(%) Energies 0, 6 45 Figure 4. Simulaion resuls of channel saus. (a) Simulaion resuls of channel conflic; (b) Simulaion resuls of channel uilisaion. 0.8 ACRA CSP ACRA CSP Simulaion Time(s) (a) Simulaion Time(s) (b) 5.. Experimen and Analysis In order o evaluae he performance of he ACRA, we designed a es-bed in our laboraory of which he environmen is similar o office or home environmens. The erminal node is shown in Figure 5. The node basically has four pars: serial communicaion par, conrol core, power line communicaion core, and coupling circui par. The serial communicaion par suppors downloading, debugging he program, and connecing wih PC. The conrol core has a Mega8 essor which conrols he operaive mode of he PLC core. The PLC core has a ST758 which suppors for muli-carrier modulaion. The coupling circui par is responsible for coupling he carrier signal o 0 V power line. Figure 5. PLC erminal node. Power line communicaion core Serial communicaion par Coupling circui par Conrol core The opology of es-bed is shown in Figure 6. I conains 0 erminal nodes. Among hem, node is he BS, which connecs wih PC via serial communicaion par o record he daa from he oher erminal nodes. The es-bed also conains compuers, air-condiioner, moor and some oher devices which are commonly used in he laboraory (e.g., lamps, scanner, priner, ec.) as noise sources. The ACRA is compared wih CSP. The accuracy rae of receiving daa is used as performance measure.

16 Accuracy rae(%) Energies 0, 6 46 In order o calculae and observe he experimenal resuls convenienly, we se ha node-node0 send characer daa a o n, or hex 0x6 o 0x6E o BS. The PC conneced wih BS is responsible for displaying and recording he daa from hese nodes. In ACRA, noe-node0 consiues a single arificial web nework, and he node6 is he cenral node. In he ess of experimenal, ACRA and CSP coninuous work 5 hours before and afer he adding of he same noise sources respecively. We record he daa and do some saisical analysis. The resuls of saisical analysis are shown in Figure 7. When here are no noise sources in he nework, he accuracy rae of CSP and ACRA is 89.87% and 99.8%, respecively. When here are similar noise sources in he nework, he accuracy rae of CSP and ACRA is 97.67% and 49.6%, respecively. Subjec o he effecs of noise, he accuracy rae of CSP and ACRA are a cerain degree of reducion, bu he ACRA is less affeced. These indicae ha because of he erminal nodes of CSP send daa disorderly, he uilizaion and conflic of channel are undesirabiliy which resul in he high bi error rae and low accuracy rae, low ani-inerference abiliy and low reliabiliy of PLC. The erminal nodes of ACRA send daa o he cenral node according o he iming in Figure 7a and Figure 4a, which avoids he conflic, improves he uilizaion rae of channel, reduces he bi error rae, ensures he accuracy rae of daa. Because of choosing he proper relay node (node 6), ACRA reduce he influence of noise on PLC effecively, improve he reliabiliy of PLC. Figure 6. Topology of he es-bed. air-condiioner 5 7 air-condiioner PC moor 9 Figure 7. Saisical resuls of he experimenal daa CSP ACRA No noise noise 6. Conclusions An arificial cobweb rouing algorihm for rouing he ree-ype physical opology of a low-volage disribuion nework in a smar grid is analysed in his paper. The arificial cobweb rouing algorihm is furher improved. Communicaion delay and hroughpu are calculaed and simulaed. The conclusions are as follows:

17 Energies 0, The novel rouing algorihm solves he problem in which some nodes canno access he PLC sysem as a resul of limied daa conneciviy and improves he reliabiliy of communicaion. Using he algorihm, he ess of roue mainenance afer neworking is simple and efficien. A rouing reconsrucion algorihm reroues bad nodes a he local nework level, avoiding he need o reroue all he nodes in he sysem and hus improving he efficiency of he communicaion sysem and enhancing he ani-desroying abiliy of he PLC sysem.. QoS in PLC is guaraneed by he communicaion delay and daa hroughpu of he arificial cobweb neworking srucure in erms of he daa essing capaciy of nodes. This rouing algorihm has he characerisics of pracicabiliy and manoeuvrabiliy.. Each node sends daa according o a sric ime series, which reduces he channel collision rae considerably and improves he channel uilisaion. Therefore, he algorihm improves he communicaion reliabiliy of he sysem. The resuls of his sudy prove ha he cobweb srucure is an effecive rouing mehod for improving he reliabiliy of LVPLC and guaraneeing QoS in PLC. This rouing algorihm may also provide a novel rouing mehod for oher neworks, such as wireless sensor neworks and ad hoc neworks. Acknowledgmens This work was suppored by he Naional Naural Science Foundaion of China (No , ). References. Mingyue, Z. Transmission characerisics of low-volage disribuion neworks in china under he smar grids environmen. IEEE Trans. Power Deliv. 0, 6, Theofilos, A.P.; Grigoris, K.P.; Peros, S.D. Low-volage disribuion line performance evaluaion for PLC signal ransmission. IEEE Trans. Power Deliv. 008,, Okazima, N.; Baba, Y.; Nagaoka, N.; Ameani, A.; Temma, K.; Shimomura, T. Propagaion characerisics of power line communicaion signals along a power cable having semiconducing layers. IEEE Trans. Elecromagn. Compa. 00, 5, Manfred, Z.; Klaus, D. Analysis and modeling of impulsive noise in broad-band powerline communicaion. IEEE Trans. Elecromagn. Compa. 00, 44, Jusinian, A.; Nelson, T.; Rajeev, T. Channel characerizaion for indoor power-line neworks. IEEE Trans. Power Deliv. 009, 4, Marco, R.; Mauro, T. Power-line communicaions channel esimaion and racking by a compeiive neural nework. IEEE Trans. Consum. Elecron. 006, 5, Dubravko, S.; Alen, B.; Roman, M. Signal propagaion modeling in power-line communicaion neworks. IEEE Trans. Power Deliv. 005, 0, Yocheol, K.; Jung, N.B.; Jin, Y.K. Performance of power line communicaion sysems wih noise reducion scheme for smar grid applicaions. IEEE Trans. Consum. Elecron. 0, 57, Julian, M. Noise analysis of power-line communicaions using spread-specrum modulaion. IEEE Trans. Power Deliv. 007,,

18 Energies 0, Ma, Y.H.; So, P.L.; Gunawan, E. Performance analysis of OFDM sysems for broadband power line communicaions under impulsive noise and mulipah effecs. IEEE Trans. Power Deliv. 005, 0, Jongman, H.; Kamrok, L.; Hyoung, K.K.; Dong-Sung, K.; Wook, H.K. Adapive channel sae rouing for home nework sysems. IEEE Trans. Consum. Elecron. 007, 5, Gao, Q.; Yu, J.Y.; Chong, P.H.J.; So, P.L.; Gunawan, E. Soluions for he silen node problem in an auomaic meer reading sysem using power-line communicaions. IEEE Trans. Power Deliv. 008,, Xiaosheng, L.; Liang, Z.; Yan, Z.; Dianguo, X. Performance Analysis of Power Line Communicaion Nework Model Based on Spider Web. In Proceedings of he IEEE 8h Inernaional Conference on Power Elecronics (ECCE Asia), Jeju, Korea, 0 May June 0; pp Sivaneasan, B.; Gunawan, E.; So, P.L. Modeling and performance analysis of auomaic meer-reading sysems using plc under impulsive noise inerference. IEEE Trans. Power Deliv. 00, 5, Charles, J.K.; Mohamed, F.C. Aenuaion characerisics of high rae home-neworking PLC signals. IEEE Trans. Power Deliv. 00, 7, Qi, J.-J.; Liu, X.-S.; Xu, D.-G.; Li, Y.; Mou, Y.-F. Simulaion sudy on cluser-based rouing algorihm and reconsrucion mehod of power line communicaion over lower-volage disribuion. Proc. CSEE 008, 8, Liu, X.-S.; Zhou, Y.; Qi, J.-J. Mehod sudy of auomaic rouing for power line communicaion. Proc. CSEE 006, 6, Konnex (KNX) sandard. KNX Associaion: Brussels, Belgium, by he auhors; licensee MDPI, Basel, Swizerland. This aricle is an open access aricle disribued under he erms and condiions of he Creaive Commons Aribuion license (hp://creaivecommons.org/licenses/by/.0/).