Pub Date : 2014-12-07DOI: 10.1109/COMSNETS.2015.7098695
Vikram P. Munishwar, Vinay Kolar, P. Jayachandran, Ravi Kokku
We present a novel efficient adaptive sensing and monitoring solution for a system of mobile sensing devices that support traffic monitoring applications. We make a key observation that much of the variance in commute times arises at a few congestion hotspots, and a reliable estimate of congestion can be obtained by selectively monitoring congestion just at these hotspots. We design a smartphone application and a back-end system that automatically identifies and monitors congestion hotspots. The solution has low resource footprint in terms of both battery usage on the sensing devices and the network bytes used for uploading data. When a user is not inside any hotspot zone, adaptive sampling conserves battery power and reduces network usage, while ensuring that any new hotspots can be effectively identified. Our results show that our application consumes 40- 80% less energy than a periodic sampling system for different routes in our experiments, with similar accuracy of congestion information. The system can be used for a variety of applications such as automatic congestion alerts to users approaching hotspots, reliable end-to-end commute time estimates and effective alternate route suggestions.
{"title":"RTChoke: Efficient real-time traffic chokepoint detection and monitoring","authors":"Vikram P. Munishwar, Vinay Kolar, P. Jayachandran, Ravi Kokku","doi":"10.1109/COMSNETS.2015.7098695","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098695","url":null,"abstract":"We present a novel efficient adaptive sensing and monitoring solution for a system of mobile sensing devices that support traffic monitoring applications. We make a key observation that much of the variance in commute times arises at a few congestion hotspots, and a reliable estimate of congestion can be obtained by selectively monitoring congestion just at these hotspots. We design a smartphone application and a back-end system that automatically identifies and monitors congestion hotspots. The solution has low resource footprint in terms of both battery usage on the sensing devices and the network bytes used for uploading data. When a user is not inside any hotspot zone, adaptive sampling conserves battery power and reduces network usage, while ensuring that any new hotspots can be effectively identified. Our results show that our application consumes 40- 80% less energy than a periodic sampling system for different routes in our experiments, with similar accuracy of congestion information. The system can be used for a variety of applications such as automatic congestion alerts to users approaching hotspots, reliable end-to-end commute time estimates and effective alternate route suggestions.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126875360","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-11-23DOI: 10.1109/COMSNETS.2015.7098709
Naveen Gupta, Anurag Singh, H. Cherifi
Understanding the epidemic dynamics, and finding out efficient techniques to control it, is a challenging issue. A lot of research has been done on targeted immunization strategies, exploiting various global network topological properties. However, in practice, information about the global structure of the contact network may not be available. Therefore, immunization strategies that can deal with a limited knowledge of the network structure are required. In this paper, we propose targeted immunization strategies that require information only at the community level. Results of our investigations on the SIR epidemiological model, using a realistic synthetic benchmark with controlled community structure, show that the community structure plays an important role in the epidemic dynamics. An extensive comparative evaluation demonstrates that the proposed strategies are as efficient as the most influential global centrality based immunization strategies, despite the fact that they use a limited amount of information. Furthermore, they outperform alternative local strategies, which are agnostic about the network structure, and make decisions based on random walks.
{"title":"Community-based immunization strategies for epidemic control","authors":"Naveen Gupta, Anurag Singh, H. Cherifi","doi":"10.1109/COMSNETS.2015.7098709","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098709","url":null,"abstract":"Understanding the epidemic dynamics, and finding out efficient techniques to control it, is a challenging issue. A lot of research has been done on targeted immunization strategies, exploiting various global network topological properties. However, in practice, information about the global structure of the contact network may not be available. Therefore, immunization strategies that can deal with a limited knowledge of the network structure are required. In this paper, we propose targeted immunization strategies that require information only at the community level. Results of our investigations on the SIR epidemiological model, using a realistic synthetic benchmark with controlled community structure, show that the community structure plays an important role in the epidemic dynamics. An extensive comparative evaluation demonstrates that the proposed strategies are as efficient as the most influential global centrality based immunization strategies, despite the fact that they use a limited amount of information. Furthermore, they outperform alternative local strategies, which are agnostic about the network structure, and make decisions based on random walks.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"142 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122727292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this paper, we propose a connection model for formation of networks having both positive and negative links. We define utility functions of a node and hence of a network by using balanced and unbalanced triads in the network. Further, using those utility functions we define strongly efficient and conditionally efficient networks. We also find sufficient conditions for which a network will be strongly/conditionally efficient. Further, we introduce pairwise stable networks having both positive and negative links. Finally, we provide certain interesting results regarding pairwise stable networks.
{"title":"Strategic network formation involving social relations: Enmity and friendship","authors":"Debabrata Pal, Bibhas Adhikari, Mainak Mazumdar","doi":"10.2139/ssrn.2532101","DOIUrl":"https://doi.org/10.2139/ssrn.2532101","url":null,"abstract":"In this paper, we propose a connection model for formation of networks having both positive and negative links. We define utility functions of a node and hence of a network by using balanced and unbalanced triads in the network. Further, using those utility functions we define strongly efficient and conditionally efficient networks. We also find sufficient conditions for which a network will be strongly/conditionally efficient. Further, we introduce pairwise stable networks having both positive and negative links. Finally, we provide certain interesting results regarding pairwise stable networks.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117030587","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2014-09-04DOI: 10.1109/COMSNETS.2015.7098691
Ratnesh Kumbhkar, M. Islam, N. Mandayam, I. Seskar
The penetration of wireless broadband services in remote areas has primarily been limited due to the lack of economic incentives that service providers encounter in sparsely populated areas. Besides, wireless backhaul links like satellite and microwave are either expensive or require strict line of sight communication making them unattractive. TV white space channels with their desirable radio propagation characteristics can provide an excellent alternative for engineering backhaul networks in areas that lack abundant infrastructure. Specifically, TV white space channels can provide “free wireless backhaul pipes” to transport aggregated traffic from broadband sources to fiber access points. In this paper, we investigate the feasibility of multi-hop wireless backhaul in the available white space channels by using noncontiguous Orthogonal Frequency Division Multiple Access (NC-OFDMA) transmissions between fixed backhaul towers. Specifically, we consider joint power control, scheduling and routing strategies to maximize the minimum rate across broadband towers in the network. Depending on the population density and traffic demands of the location under consideration, we discuss the suitable choice of cell size for the backhaul network. Using the example of available TV white space channels in Wichita, Kansas (a small city located in central USA), we provide illustrative numerical examples for designing such wireless backhaul network.
{"title":"Rate optimal design of a wireless backhaul network using TV white space","authors":"Ratnesh Kumbhkar, M. Islam, N. Mandayam, I. Seskar","doi":"10.1109/COMSNETS.2015.7098691","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098691","url":null,"abstract":"The penetration of wireless broadband services in remote areas has primarily been limited due to the lack of economic incentives that service providers encounter in sparsely populated areas. Besides, wireless backhaul links like satellite and microwave are either expensive or require strict line of sight communication making them unattractive. TV white space channels with their desirable radio propagation characteristics can provide an excellent alternative for engineering backhaul networks in areas that lack abundant infrastructure. Specifically, TV white space channels can provide “free wireless backhaul pipes” to transport aggregated traffic from broadband sources to fiber access points. In this paper, we investigate the feasibility of multi-hop wireless backhaul in the available white space channels by using noncontiguous Orthogonal Frequency Division Multiple Access (NC-OFDMA) transmissions between fixed backhaul towers. Specifically, we consider joint power control, scheduling and routing strategies to maximize the minimum rate across broadband towers in the network. Depending on the population density and traffic demands of the location under consideration, we discuss the suitable choice of cell size for the backhaul network. Using the example of available TV white space channels in Wichita, Kansas (a small city located in central USA), we provide illustrative numerical examples for designing such wireless backhaul network.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115444846","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098727
Ayush Kumar, Chaitanya Kansal, Asif Ekbal
Active Learning is a technique to automatically select the useful instances from the unlabelled data in such a way that, when these are augmented to the training data, overall classification performance improves. The creation of training examples otherwise involves significant amount of costs and efforts and hence, is a major constraint in the supervised algorithms. In this paper, we investigate the effectiveness of active learning for sentiment classification of Tweets. The algorithm selects the informative unlabelled data based on the concept of uncertainty sampling which dictates that only those Tweets be added to the training set for which the classifier can quickly refine its decision boundary. Our experiments on a benchmark dataset of Tweets show an overall accuracy of 83.95%, which is an increment of 6.75% over the baseline model, constructed by training a Support Vector Machine (SVM) with all the available set of features. The approach, being very general, is scalable, domain-adaptable and easy to implement for a wide variety of problems.
{"title":"Investigating active learning techniques for document level sentiment classification of tweets","authors":"Ayush Kumar, Chaitanya Kansal, Asif Ekbal","doi":"10.1109/COMSNETS.2015.7098727","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098727","url":null,"abstract":"Active Learning is a technique to automatically select the useful instances from the unlabelled data in such a way that, when these are augmented to the training data, overall classification performance improves. The creation of training examples otherwise involves significant amount of costs and efforts and hence, is a major constraint in the supervised algorithms. In this paper, we investigate the effectiveness of active learning for sentiment classification of Tweets. The algorithm selects the informative unlabelled data based on the concept of uncertainty sampling which dictates that only those Tweets be added to the training set for which the classifier can quickly refine its decision boundary. Our experiments on a benchmark dataset of Tweets show an overall accuracy of 83.95%, which is an increment of 6.75% over the baseline model, constructed by training a Support Vector Machine (SVM) with all the available set of features. The approach, being very general, is scalable, domain-adaptable and easy to implement for a wide variety of problems.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125573232","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098711
Harry Raymond Joseph, G. Raina, K. Jagannathan
Motivated by concerns due to rapidly growing vehicular congestion in Indian cities, we conduct a preliminary investigation into the economic costs of congestion in Delhi. In particular, we estimate the marginal and the total costs of congestion in Delhi. In estimating the marginal costs, we consider the following factors: (i) productivity loss, (ii) air pollution costs, and (iii) costs due to accidents. In calculating the total costs, in addition to the above factors, we also estimate the costs due to the wastage of fuel. We also project the associated costs due to productivity loss and air pollution till 2030. The key takeaway from our current study is that costs due to productivity loss, particularly from buses, dominates the overall economic costs. Additionally, the expected increase in fuel wastage makes a strong case for intelligent traffic management systems.
{"title":"Cost estimates for road congestion in Delhi: projections and recommendations","authors":"Harry Raymond Joseph, G. Raina, K. Jagannathan","doi":"10.1109/COMSNETS.2015.7098711","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098711","url":null,"abstract":"Motivated by concerns due to rapidly growing vehicular congestion in Indian cities, we conduct a preliminary investigation into the economic costs of congestion in Delhi. In particular, we estimate the marginal and the total costs of congestion in Delhi. In estimating the marginal costs, we consider the following factors: (i) productivity loss, (ii) air pollution costs, and (iii) costs due to accidents. In calculating the total costs, in addition to the above factors, we also estimate the costs due to the wastage of fuel. We also project the associated costs due to productivity loss and air pollution till 2030. The key takeaway from our current study is that costs due to productivity loss, particularly from buses, dominates the overall economic costs. Additionally, the expected increase in fuel wastage makes a strong case for intelligent traffic management systems.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114739777","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098702
S. Mukhopadhyay, M. Pramod, Anurag Kumar
Heterogeneous traffic consisting of medium sized cars and two-wheeled vehicles, such as motorcycles, arrive at a single-lane leg of a signalized road intersection. The lane is controlled by periodic “green” and “red” periods. The traffic is lane indisciplined in that, instead of standing one behind the other, the two-wheelers fill up the lane width-wise, by standing side-to-side with each other or with the cars. This gives rise to a queueing model in which the vehicles form batches (e.g., up to four motorcycles side-to-side in a batch, or a car and up to two motorcycles side-to-side) and each batch exits the intersection together. Assuming a Poisson point process model for vehicle arrivals, we approximately analyze this interrupted queue system by viewing it as an assembly queue followed by an interrupted M/SM/1 queue (where SM stands for semi-Markov). Analysis of the assembly queue provides a Markov model for the types of the successive batches in the intersection, and thereby characterizes the semi-Markov process of service times. The mean delay in the interrupted M/SM/1 queue is approximately analyzed by employing an extension of the Webster mean delay formula. Numerical results are provided to illustrate how well the approximation works in several examples.
{"title":"An approach for analysis of mean delay at a signalized intersection with indisciplined traffic","authors":"S. Mukhopadhyay, M. Pramod, Anurag Kumar","doi":"10.1109/COMSNETS.2015.7098702","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098702","url":null,"abstract":"Heterogeneous traffic consisting of medium sized cars and two-wheeled vehicles, such as motorcycles, arrive at a single-lane leg of a signalized road intersection. The lane is controlled by periodic “green” and “red” periods. The traffic is lane indisciplined in that, instead of standing one behind the other, the two-wheelers fill up the lane width-wise, by standing side-to-side with each other or with the cars. This gives rise to a queueing model in which the vehicles form batches (e.g., up to four motorcycles side-to-side in a batch, or a car and up to two motorcycles side-to-side) and each batch exits the intersection together. Assuming a Poisson point process model for vehicle arrivals, we approximately analyze this interrupted queue system by viewing it as an assembly queue followed by an interrupted M/SM/1 queue (where SM stands for semi-Markov). Analysis of the assembly queue provides a Markov model for the types of the successive batches in the intersection, and thereby characterizes the semi-Markov process of service times. The mean delay in the interrupted M/SM/1 queue is approximately analyzed by employing an extension of the Webster mean delay formula. Numerical results are provided to illustrate how well the approximation works in several examples.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"344 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124313902","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098680
Abhishek K. Gupta, Kushal Bansal
In window-based flow control mechanism for Content Centric Networking (CCN), for one Interest (a content request packet), more than one Content Objects (content chunks) are expected to be received. TCP's RTT estimation algorithm cannot be used directly for such scenarios because for some contents, there will be no Interest reference. In this paper, we extend TCP's RTT estimation for window-based flow control mechanism. We also estimate the processing time delay and use the estimate to compute a number of timer values. The proposed mechanism is supported by various experimental results. In addition to that, we also propose an optimization for Content Store processing in window-based flow control mechanism for the received Interest packet and compare the results obtained from the actual implementation.
{"title":"Flow control enhancements in Content Centric Networking","authors":"Abhishek K. Gupta, Kushal Bansal","doi":"10.1109/COMSNETS.2015.7098680","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098680","url":null,"abstract":"In window-based flow control mechanism for Content Centric Networking (CCN), for one Interest (a content request packet), more than one Content Objects (content chunks) are expected to be received. TCP's RTT estimation algorithm cannot be used directly for such scenarios because for some contents, there will be no Interest reference. In this paper, we extend TCP's RTT estimation for window-based flow control mechanism. We also estimate the processing time delay and use the estimate to compute a number of timer values. The proposed mechanism is supported by various experimental results. In addition to that, we also propose an optimization for Content Store processing in window-based flow control mechanism for the received Interest packet and compare the results obtained from the actual implementation.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124465462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098724
Anirudh Vemula, Nikhil Patil, Vivek Paharia, A. Bansal, Megha Chaudhary, N. Aggarwal, D. Bansal, K. Ramakrishnan, B. Raman
Commuting on roads in densely populated cities of the developing world is fraught with high delays and uncertainties. Wide use of public transportation can ease the load on the road infrastructure, but such use is not convenient, partly due to the unpredictable nature. In this work, our goal is to improve the usability of public transportation, through better information. Such information can lead to better planning and predictability for commuters. We take a crowd-sourced approach where information about transportation units as well as road conditions is crowd-sourced from commuters. The information is then processed and made available to other commuters. In this context, this paper presents a naming framework we have developed, which will enable flexible and scalable content-driven data gathering and dissemination. Based on a preliminary implementation of the framework, we present various field-experiment results which shed light on the practicality of the proposed approach as well as on technical issues which need further careful addressing.
{"title":"Improving public transportation through crowd-sourcing","authors":"Anirudh Vemula, Nikhil Patil, Vivek Paharia, A. Bansal, Megha Chaudhary, N. Aggarwal, D. Bansal, K. Ramakrishnan, B. Raman","doi":"10.1109/COMSNETS.2015.7098724","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098724","url":null,"abstract":"Commuting on roads in densely populated cities of the developing world is fraught with high delays and uncertainties. Wide use of public transportation can ease the load on the road infrastructure, but such use is not convenient, partly due to the unpredictable nature. In this work, our goal is to improve the usability of public transportation, through better information. Such information can lead to better planning and predictability for commuters. We take a crowd-sourced approach where information about transportation units as well as road conditions is crowd-sourced from commuters. The information is then processed and made available to other commuters. In this context, this paper presents a naming framework we have developed, which will enable flexible and scalable content-driven data gathering and dissemination. Based on a preliminary implementation of the framework, we present various field-experiment results which shed light on the practicality of the proposed approach as well as on technical issues which need further careful addressing.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114829286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/COMSNETS.2015.7098712
K. J. Prabuchandran, Hemanth Kumar A.N, S. Bhatnagar
In this paper, we study the problem of obtaining the optimal order of the phase sequence [14] in a road network for efficiently managing the traffic flow. We model this problem as a Markov decision process (MDP). This problem is hard to solve when simultaneously considering all the junctions in the road network. So, we propose a decentralized multi-agent reinforcement learning (MARL) algorithm for solving this problem by considering each junction in the road network as a separate agent (controller). Each agent optimizes the order of the phase sequence using Q-learning with either ∈-greedy or UCB [3] based exploration strategies. The coordination between the junctions is achieved based on the cost feedback signal received from the neighbouring junctions. The learning algorithm for each agent updates the Q-factors using this feedback signal. We show through simulations over VISSIM that our algorithms perform significantly better than the standard fixed signal timing (FST), the saturation balancing (SAT) [14] and the round-robin multi-agent reinforcement learning algorithms [11] over two real road networks.
{"title":"Decentralized learning for traffic signal control","authors":"K. J. Prabuchandran, Hemanth Kumar A.N, S. Bhatnagar","doi":"10.1109/COMSNETS.2015.7098712","DOIUrl":"https://doi.org/10.1109/COMSNETS.2015.7098712","url":null,"abstract":"In this paper, we study the problem of obtaining the optimal order of the phase sequence [14] in a road network for efficiently managing the traffic flow. We model this problem as a Markov decision process (MDP). This problem is hard to solve when simultaneously considering all the junctions in the road network. So, we propose a decentralized multi-agent reinforcement learning (MARL) algorithm for solving this problem by considering each junction in the road network as a separate agent (controller). Each agent optimizes the order of the phase sequence using Q-learning with either ∈-greedy or UCB [3] based exploration strategies. The coordination between the junctions is achieved based on the cost feedback signal received from the neighbouring junctions. The learning algorithm for each agent updates the Q-factors using this feedback signal. We show through simulations over VISSIM that our algorithms perform significantly better than the standard fixed signal timing (FST), the saturation balancing (SAT) [14] and the round-robin multi-agent reinforcement learning algorithms [11] over two real road networks.","PeriodicalId":277593,"journal":{"name":"2015 7th International Conference on Communication Systems and Networks (COMSNETS)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123052571","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：