Communiations and Networ, 2013, 5, 69-73 http://dx.doi.org/10.4236/n.2013.53b2014 Published Online September 2013 (http://www.sirp.org/journal/n) Cross-layer Resoure Alloation on Broadband Power Line Based on Novel QoS-priority Sheduling Funtion in MAC Layer Huang Qian, Lu Jun, Xiong Chen, Duan Ruihao Shool of Eletrial and Eletroni Engineering, North China Eletri Power University, Beijing, China Email: hq050176@163.om, lujun@nepu.edu.n, xionghen2007@163.om, duanr501@163.om Reeived May, 2013 ABSTRACT Traditional resoure alloation algorithms use the hierarhial system, whih does not apply to the bad hannel environment in broadband power line ommuniation system. Introduing the idea of ross-layer an improve the utilization of resoures and ensure the QoS of servies. This paper proposes a ross-layer resoure alloation on broadband power line based on QoS priority sheduling funtion on MAC layer. Firstly, the algorithm onsiders both of real-time users requirements for delay and non-real-time users requirements for queue length. And then user priority funtion is proposed. Then eah user s sheduled paets number is alulated aording to its priority funtion. The sheduling sequenes are based on the utility funtion. In physial layer, aording to the sheduled paets, the algorithm alloates physial resoures for paets. The simulation results show that the proposed algorithm give onsideration to both lateny and throughput of the system with improving users QoS. Keywords: Broadband Power Line Communiation; OFDM; Cross-layer; Resoure Alloation; Sheduling 1. Introdution With the onstrution of strong and smart grid and the rapid development of ommuniation tehnology, broadband power line ommuniations gradually develop into the high-speed. Traditional wired networ The wired networs adopt traditional hierarhial system: Open Systems Interonnetion (OSI) model, whih, from top to bottom, is divided into the appliation layer, presentation layer, session layer, transport layer, networ layer, data lin layer, and physial layer. Eah layer is independent designed, and the interfae between the layers is stati. In broadband power line systems, due to the large differenes of sub-hannels, great hanges of noise, multipath fading, dispersion and other unfavorable fators, the traditional hierarhial system annot be flexible to adapt to fast-hanging hannel environment. Apply the ross-layer design into the OFDM system, and let the upper layer s dynami business be in lose ontat with the physial layer s hannel onditions, whih ould realize the dynami alloation of resoures and further improve the utilization of resoures. In referene [1], an algorithm based on the satisfation of paet exhange and resoure alloation is proposed. The algorithm optimally defines the various satisfatory degree evaluation funtions. Then, based on suh funtions, the algorithm alulates sheduling utility funtion, ahieving user s sheduling. In this algorithm, the differentiation between emergeny servies and non-emergeny servies is not obvious, whih may lead to paet loss of real-time business. In referene [2], a sheduling algorithm based on the satisfatory fator is proposed. The algorithm defines satisfatory fator of quality of servie, and gets the new utility funtion through improving the traditional index sheduling algorithm (EXP). The algorithm does not onsider the queue s onditions, whih ould mae paet loss rate inrease for non-real-time servies. This paper introdues a ross-layer resoure alloation algorithm on broadband power line based on a novel sheduling funtion in MAC layer. For real-time business, the algorithm onsiders the paets delay; for non-real-time business, it onsiders the queue length. Combining with the user hannel onditions and fators above, users are divided into different priorities. And MAC layer utility funtion is proposed. In physial layer, we use equal power resoure alloation method. The algorithm losely ontats the MAC layer with the physial layer, and proesses paet sheduling and resoure alloation reasonably, whih improves the throughput of the system. 2. Model of Cross-layer Resoure Alloation Through information exhanging between layers, and
70 H. QIAN ET AL. aording to the system s overall onditions, the rosslayer resoure alloation algorithm meets uses needs better in eah layer. Therefore, ross-layer alloation algorithm provides more information to adaptive resoure alloation. However, beause of the inrease of information, omputational omplexity of the algorithm will inrease [3]. The model of ross-layer resoure alloation is shown below: Aording to the information interation between layers and relevant sheduling rules, the sheduler reasonably alloates system resoures, as is shown in Figure 1. The information between layers mainly inludes hannel estimation, feedba, resoures of physial layer, queue status and the QoS requirements, et. In broadband power line system, MAC-PHY ross-layer sheduler wors as follows: data from the upper layer, whih is assigned to different buffer queues, ombines with physial hannel onditions as the basis of the MAC layer sheduler. In the broadband power line system, the operations of the upper layer are varied, and the QoS requirements are also different. In addition, the arrival of some business flows is sudden and random, and the power line hannel is time-varying. To mae full use of lin resoures, and ensure the validity and reliability, the sheduler must tae real-time hannel status into onsideration. Thus, the sheduler ould selet the MAC-PHY ross-layer sheduling mode above. 3. Cross-layer Sheduling and Resoure Alloation Algorithm 3.1. User Sheduling on MAC The sheduling is implemented in the MAC layer. The utility funtion may be provided on the joint of the delay, queue length, paet loss rate and other information on the MAC layer and hannel information on the physial layer. The utility funtion determines the sheduling order of business, and ompletes sheduling [4]. For the sae of simpliity, it is assumed that, at the same moment, a user has only one servie, and orresponds to only one queue. Eah data paet sheduling is for eah queue sheduling. Corresponds to the real-time and nonreal-time user, the queue is divided into real-time queue and non-real-time one. First, aording to the urgeny degree of every queue, paets are divided into two ategories: high priority paets and ordinary paets. The priority of queue of real-time business is determined by delay. Let maximum delay whih an be tolermax ated by real-time business of user is D and assume waiting time of paets of real-time business queue of urrent user is D, the priority of paets of real-time business queue of user is divided aording to equation (1). D RT (1) D urrent max Paet loss of non-real-time business queue is mainly aused by the overflow of paets, so its priority is determined by the length of queue [5]. Set the maximum number of buffered paets of buffered queue is Q max. In the slot t, the prioritization funtion of non-real-time business queue of user is: Q NRT (2) Qmax where is the oordination fator, whih is a positive number less than 1. The emergeny degree of data paets in eah queue is determined as follows: 1high priority (3) ordinary where shows the emergeny funtion of real-time or non-real-time servies, is the threshold of emergeny funtion. When the priority of the head paet of eah queue is judged, sheduling sequenes of different priority is judged by the sheduling funtion of high priority and ordinary paets. Sheduling funtion is expressed as follows: PD [] i [ i]high priority PD U [] K i (4) R [] i [ i] ordinary a R [] i where PD [] i represents the paet loss rate of the queue before the i-th frame, and PD K represents the maximum paet loss rate. And R [] i represents the maximum supported rate of the queue s i-th frame, that is, when all of time-frequeny blos of the i-th frame distribute to the queue, the transmission rate an reah the maximum [6]. And R a [] i represents the weighted average rate of the queue before the i-th frame. The iterative formula desribes as follows: 1 1 a a R[] i (1 ) R[ i 1] R[ i 1] t t (5) where R [ i 1] shows the atual rate of the i-1 frame of queue, and t is the length of the sliding time window. Based on the sheduling funtion above, paets of eah queue are sheduled. The sheduling priniple is that, dispath high priority paets firstly, and then, shedule ordinary paets. The paets of same priority are sheduled aording to the desending order of sheduling funtion. Average sending rate of eah user is indiated as a
H. QIAN ET AL. 71 vetor: Paet Rate [ R1, R2,, R K ]. Aording to the priority funtion, the number of urrent sheduling paets of eah user is determined as follows. queue N R (1 ) (6) where is the normalized priority. (7) K 1 3.2. Physial Resoure Alloation The method of equal power alloation is applied on physial layer. Suppose the total power of the system is P, the number of subarriers is N, the number of bits whih an be transmitted for user on the subarrier n is: Rb ( P/ N) gn, log (1 ) n, 2 In aordane with the number of sheduling paets identified on the MAC layer, the sub-arrier whih is the optimal mathing is assigned to eah user. When the resoure alloation on physial layer is omplete, the urrent round of resoure alloation is end. If the rate of users reahes the sheduling requirement and the system also has the remaining resoures, the remaining resoures are needed to be alloated. 3.3. Remaining Resoure Alloation If all users have reahed the fixed rate and the unused available sub-arriers are still present in the system, the algorithm alloates the remaining resoures. Remaining resoure alloate priniple is as follows. Export the highest priority paet aording to the MAC layer sheduling algorithm. The optimal sub-arrier is assigned to orresponding paet until bits alloation of the paet is ompleted. Figure 1. Cross-layer resoure alloation system diagram. (8) Figure 2. The flowhart of ross-layer resoure alloation. If the system does not have enough sub-arriers to transmit bits of the paets, they are not sent in this OFDM symbol, and the alloation algorithm of the present OFDM symbol finishes. If the alloation an be finished and the system still has sub-arriers to be used, return to the first step. The flowhart of ross-layer resoure alloation on broadband power line ommuniation is shown in Figure 2. 4. Simulation and Analysis 4.1. Traffi Soure Model With the development of broadband power line ommuniation, servies hosted on it beome diverse inreasingly. In order to failitate the analysis, servie whih has high time delay requirements belongs to real-time servie, and others are non-real-time servie. Two types of servies are modeled. Paets of real-time servie queue arrive obey the twostate Marov distribution, as is shown in Figure 3. ON indiates the ativation period. OFF indiates the quiet period. and indiate the orresponding transition probability.
72 H. QIAN ET AL. 1 ( T / T OFF ) (9) 1exp( T / T ON ) (10) In whih T is the length of one OFDM symbol, TOFF is the average of quiet period, T ON is the average of ativation period. Paets of non-real-time servie queue arrive obey the Poisson distribution. et( t) n PN ( n) (11) n! In whih is the average paets arrival rate of nonreal-time servie. In order to ensure the stability of the system, the average apaity of servies should not exeed the system s total transmit apaity. Servie soure models reated using above distributions are shown in Figure 4 and Figure 5. 1.5 Figure 3. The real-time servie Marov model. Two-state Marov 4.2. Simulation and Analysis of System Performane The algorithm simulates in typial power line environment. The following parameters are set as follows: The frequeny range is 0-20 MHZ, power spetrum upper is -50-0.8 f (dbm/hz) and f is in MHz. The number of sub-arriers is 128. Real-time business is orresponding to user 1 and 2 and the maximum delay is 10ms. Its maximum paet loss rate is 10-3. Non-real-time business is orresponding to user 3 and 4 and its maximum queue length is 600. Its maximum paet loss rate is 10-3. = 0.7. Four users average paets arrival rates are: 60 bps, 60 bps, 40 bps and 30 bps. This artile selets maximum throughput algorithm [7] (MAX/MIN) as the omparison. And the proposed algorithm ompares and analyses the following fators, suh as system throughput, delay and paet loss rate. Throughput performane omparison Figure 6 depits that the MAX / MIN has an uneven distribution. User one has high throughput, but other users throughput is low. This paper algorithm s throughput mathes users average paets arrival rates and has high throughput. Delay performane omparison 1 unitary data 0.5 0 0 100 200 300 400 500 600 700 800 900 1000 Time Figure 4. The real-time servie model. Figure 6. Throughput performane. 1.8 Poisson 1.6 1.4 unitary data 1.2 1 0.8 0.6 0.4 0 100 200 300 400 500 600 700 800 900 1000 Time Figure 5. The non-real-time servie model. Figure 7. Average queue length.
H. QIAN ET AL. 73 Figure 7 shows the average queue length of eah user. The MAX\MIN has long queue length. It shows that users have higher delay. This paper algorithm has short queue length, whih means the delay of paets is short. Performane omparison of paet loss rate. Statistis show that the paet loss rate of MAX / I is 0.0015. The paet loss rate of this paper algorithm is zero. The algorithm proposed taes the users average paets arrival rate into aount, improving the paets transmission rate. And, the algorithm possessed gets lower paet loss rate and ensures the user's QoS. 5. Conlusions This paper disusses ross-layer resoure alloation on broadband power line ommuniation system and puts forward a MAC layer sheduling algorithm based on users priority. The algorithm reates different sheduling funtion between the real-time and non-real-time users in MAC layer, and applies the priority funtion to divide users into two priorities. Then users are sheduled based on the priority and utility funtion. In the physial layer, equal power alloation method is used. The simulation results show that the proposed algorithm an tae the lateny and throughput into aount and improve user's QoS. 6. Anowledgements This wor was supported by National Natural Siene Foundation of China (No.51007021) and National Siene and Tehnology Major Projet of the Ministry of Siene and Tehnology of China (No.2010ZX03006-005-01). REFERENCES [1] J. Huang, G. S. Vijay and A. Rajeev, et al., Joint Sheduling and Resoure Alloation in Uplin OFDM Systems for Broadband Wireless Aess Networs, IEEE J. Selet. Areas Commun. Vol. 27, No. 2, 2009, pp. 226-234. doi:10.1109/jsac.2009.090213 [2] C. Fan, L. Zhang and W. J. Lian, A Sheduling Algorithm for Guarantying QoS of Streaming Traffi over Mixed Servies, Journal of Bijing University of Posts and Teleommuniations, Vol. 30, No. 3, 2007, pp. 75-78. [3] W. W. L. Ho and Y. C. Liang, Optimal Resoure Alloation for Multiuser MIMO-OFDM Systems with User Rate Constraints, IEEE Trans. Veh. Tehnol., Vol. 58, No. 3, 2009, pp. 1190-1203. doi:10.1109/tvt.2008.927721 [4] K. Deb, A. Pratap and S. Agarwal, A Fast and Elitist Multiobjetive Geneti Algorithm: NSGA- Ⅱ, IEEE Transations on Evolutionary Computation, Vol. 6, No. 2, 2002, pp. 182-197. [5] S. Y. Lin and J. S. Huang, Adaptive Subarrier Assignment and Bit Alloation for Multiuser OFDM System Using Ordinal Optimization Approah, IEEE Trans. Veh. Tehnol., Vol. 57, No. 5, 2008, pp. 2907-2919. [6] H. Hou, W. Zhou, S. Zhou, et al., Cross-layer Resoure Alloation for Heterogeneous Traffis in Multiuser OFDM based on a New QoS Fairness Criterion, in Pro. VTC, Baltimore, MD, USA, Sept. 2007, pp. 1593-1597. [7] B. Sanghamitra, S. Sriparna and U. Mauli, A Simulated Annealing-based Multiobjetive Optimization Algorithm: AMOSA, IEEE Transations on Evolutionary Computation, Vol. 12, No. 3, 2008, pp. 269-283.doi:10.1109/TEVC.2007.900837