Leveraging peer to peer energy sharing for resource optimization

Virginia Commonwealth University VCU Scholars Compass Theses and Dissertations Graduate School 2020 LEVERAGING PEER-TO-PEER ENERGY SHARING FOR RESOURCE OPTIMIZATION IN MOBILE SOCIAL NETWORKS Aashish Dhungana Virginia Commonwealth University Follow this and additional works at: https://scholarscompass.edu/etd Part of the Digital Communications and Networking Commons, and the Theory and Algorithms Commons © The Author Downloaded from https://scholarscompass.edu/etd/6439 This Dissertation is brought to you for free and open access by the Graduate School at VCU Scholars Compass. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of VCU Scholars Compass. For more information, please contact libcompass@vcu. c Aashish Dhungana, October 2020 All Rights Reserved.

LEVERAGING PEER-TO-PEER ENERGY SHARING FOR RESOURCE OPTIMIZATION IN MOBILE SOCIAL NETWORKS A dissertation submitted in fulfillment of the requirements for the degree of Doctor of Philosophy at Virginia Commonwealth University. by AASHISH DHUNGANA Doctorate in Computer Engineering with specialization in Computer Science - 2016-2020 Director: Dr. Eyuphan Bulut, Associate Professor, Department of Computer Science Virginia Commonwewalth University Richmond, Virginia October, 2020 i Acknowledgements First, I would like to acknowledge and express my sincere gratitude to my honorable advisor Dr. Eyuphan Bulut, for his direction, assistance, and invaluable guidance.

I would also like to thank Dr. Tomasz Arodz, Dr. Kostadin Damevski, Dr. Kemal Akkaya and Dr.

Yanxiao Zhao for their kind approval to join my dissertation committee and for providing guidance and valuable feedback. I would also like to acknowledge and thank my family for their continuous assistance during my entire academic journey. ii TABLE OF CONTENTS Chapter Page Acknowledgements. ii Table of Contents.

iii List of Tables. v List of Figures .3 Organization of the Dissertation .1 Energy Sharing Technologies .2 Mobile Social Networks .1 Optimal Energy Usage .2 Energy Sharing for Content Delivery .3 Energy Distribution for Balancing. 17 3 Mobile Charging Relief via P2P Energy Sharing .3 Dynamic Programming based Optimization .1 Optimization for Conservative Charging .2 Optimization for Cooperative Charging .4 Network-wise Optimization. 47 4 Content Delivery with P2P Energy Sharing .3 Optimal Stopping Theory .2 Energy and Residual Time-to-Live relation .3 Optimal Content and Energy Sharing .1 Algorithms in Comparison.

71 5 Energy Balancing with P2P Energy Sharing .3 Energy Balancing for Fully Connected Graphs .1 Greedy Positive First Energy Balancing (PGP ) .2 Greedy Closer First Energy Balancing (PGC ) .3 Greedy Optimal Energy Balancing (PGO ) .4 Energy Balancing for Partially Connected Graphs .1 Energy Balancing with Single Hop Energy Exchanges .1 Optimal Energy Balance .2 Energy Balancing Protocols .2 Energy Balancing with Multi-Hop Energy Exchanges .1 Optimal Energy Balance .2 Energy Balancing Protocol .1 Energy Balancing Protocols in Comparison .4 Fully Connected Graphs .5 Partially Connected Graphs .6 Discussion on Network Lifetime Maximization. 123 7 Future Research Directions. 142 v LIST OF TABLES Table Page 1 A summary of current research using energy sharing in mobile social networks. 18 2 Notations used in Chapter 3.

23 3 (Source, destination) index assignments for D matrix’s fourth dimen- sion based on charging decisions of users with different types of decision blocks. 34 4 Optimal charging decisions in each charging scenario. 40 5 Charging decisions for each decision block in cooperative case. 41 6 Notations used in Chapter 4.

57 7 Decisions with forwarding and sharing. 59 8 Simulation settings for Chapter 4. 63 9 Energy transfer amounts between nodes and final energy levels of nodes for scenarios in Fig.20 with 80% transfer efficiency. 75 10 Notations used in Chapter 5.

78 vi LIST OF FIGURES Figure Page 1 The scenarios considered for energy sharing in different mobile network applications. 4 2 Energy sharing scenarios in a mobile social network consisting of smart mobile devices. The energy sharing can be achieved in a conductive manner via a sharing cable or a gadget or through near-field wireless power transfer. 14 3 Source node charges itself at a charger and when it meets with a mes- senger offering better delivery option for its message to a specific des- tination, it transfers the message as well as the sufficient energy for the messenger to carry it to the destination.

16 4 Charging patterns and decision points of two users. 25 5 Total duration with energy exchange opportunity determined by the intersection of user meetings, charging patterns and charging decisions of users. 32 6 Dynamic programming table cell updates in the fourth dimension on a sample charging pattern of two users with different charging types included in decision blocks. 33 7 Charging patterns and skips after cooperative charging.

Arrows show the direction and the amount of energy shared between the users. 40 8 Statistics from real mobile network traces: a) distribution of number of meetings between pairs of nodes, b) hourly distribution of meeting times between nodes during a day, and c) distribution of meeting durations. 43 9 CDF of mobile charging relief obtained among all users and pairs with conservative and collaborative charging, respectively. 44 10 Average mobile charging relief with conservative and different collab- orative charging cases.

45 vii 11 Average mobile charging relief with different number of days of data used. 46 12 Impact of wireless power transfer efficiency and speed on the average mobile charging relief. 47 13 An illustration of energy sharing based content delivery in opportunis- tic networks, where energy is used as an incentive to carry a message copy. 50 14 An example opportunistic network with mean intermeeting times de- noted as the weights of the edges on the graph.

52 15 Delivery rate, delay and number of forwardings versus time-to-live in Cambridge dataset. 67 16 Delivery rate, delay and number of forwardings versus time-to-live in Haggle dataset. 68 17 Delivery rate, delay and number of forwardings versus time-to-live in synthetic dataset. 69 18 Impact of loss rate, transfer efficiency and available partial link weight on the performance ratio of sharing over forwarding.

70 19 Energy balancing through interactions between nodes at opposite sides of the average energy in the network. 73 20 (a) Energy balancing in a fully connected contact graph. (b) Energy balancing in a partially connected contact graph. (c) Energy Balancing with time limit of 50.

Edges represent that the nodes meet each other opportunistically with an average intermeeting time shown as link weight. 74 21 Optimal target average energy for different energy loss rates for a large- scale network with uniform energy distributions. 86 22 An example contact graph with 3 nodes: (a) Perfect energy balanc- ing is possible with single hop energy exchanges. (b) Perfect energy balancing requires multi-hop energy exchanges (with β = 0.

94 viii 23 Comparison of proposed algorithms with the state-of-the-art algorithm in terms of (a) variation distance, (b) total energy remaining in the network, (c) total number of interactions, (d) variation distance at each total energy level and (e) variation distance at each total number of interactions (when β=0. (f) shows the impact of different loss rates on PGO performance. 107 24 Impact of time threshold (τ ) and loss rate (β) on optimal average en- ergy achievable (Eopt ) and corresponding variation distance and total loss at Eopt with expected meeting probability threshold p = 1 − 1/e = 0. 109 25 Comparison of protocols in terms of (a) variation distance, (b) total energy remaining in the network, (c) total number of interactions, (d) variation distance at each total energy level and (e) variation distance at each total number of interactions (when β=0.63) using regular synthetic traces.

(f) shows variation distance with p=0. 111 26 Comparison of protocols in terms of (a) total energy remaining in the network, (b) variation distance at each total energy level and (c) variation distance at each total number of interactions (when β=0.63) using regular synthetic traces. (d) shows total energy remaining in the network with p=0. 113 27 Comparison of protocols in terms of (a) variation distance, (b) total energy remaining in the network, (c) total number of interactions, (d) variation distance at each total energy level and (e) variation distance at each total number of interactions (when β=0.63) using Cambridge traces.

114 28 Comparison of all algorithms in terms of (a) variation distance, (b) to- tal energy remaining in the network, (c) total number of interactions, (d) variation distance at each total energy level and (e) variation dis- tance at each total number of interactions (when β=0.8) using group-based synthetic traces. 116 ix 29 Comparison of PLE and PM LE in terms of (a) variation distance, (b) total energy remaining in the network, and (c) total number of inter- actions under different inter-group contact sparsity (γ) in group-based synthetic traces (p = 0. 118 30 Comparison of protocols in terms of achievable network lifetime with balancing and lifetime maximization objective functions and different γ values (when β=0.8) using group-based synthetic traces. 120 x Abstract LEVERAGING PEER-TO-PEER ENERGY SHARING FOR RESOURCE OPTIMIZATION IN MOBILE SOCIAL NETWORKS By Aashish Dhungana A dissertation submitted in fulfillment of the requirements for the degree of Doctor of Philosophy at Virginia Commonwealth University.

Virginia Commonwealth University, 2020. Eyuphan Bulut, Associate Professor, Department of Computer Science Mobile Opportunistic Networks (MSNs) enable the interaction of mobile users in the vicinity through various short-range wireless communication technologies (e., Bluetooth, WiFi) and let them discover and exchange information directly or in ad hoc manner. Despite their promise to enable many exciting applications, limited battery capacity of mobile devices has become the biggest impediment to these appli- cations. The recent breakthroughs in the areas of wireless power transfer (WPT) and rechargeable lithium batteries promise the use of peer-to-peer (P2P) energy sharing (i., the transfer of energy from the battery of one member of the mobile network to the battery of the another member) for the efficient utilization of scarce energy resources in the network.

However, due to uncertain mobility and communication opportunities in the network, resource optimization in these opportunistic networks is very challenging. In this dissertation, we study energy utilization in three different applications in Mobile Social Networks and target to improve the energy efficiency in the network by benefiting from P2P energy sharing among the nodes. More specifi- xi cally, we look at the problems of (i) optimal energy usage and sharing between friendly nodes in order to reduce the burden of wall-based charging, (ii) optimal content and energy sharing when energy is considered as an incentive for carrying the content for other nodes, and (iii) energy balancing among nodes for prolonging the network lifetime. We have proposed various novel protocols for the corresponding applications and have shown that they outperform the state-of-the-art solutions and improve the energy efficiency in MSNs while the application requirements are satisfied.

xii CHAPTER 1 INTRODUCTION About 5 billion users are carrying a mobile device with a service around the globe [1]. The various uses of these devices and increasing popularity of software applications such as email, Facebook, and maps have made people highly dependent on mobile devices. This intensive use of mobile devices has brought a huge load on battery re- quirements. The hardware capabilities have significantly improved since the advent of smartphones but the development of powerful batteries have not taken the necessary pace, making the batteries the main bottleneck.

The charge on most smartphones lasts about one day with average usage, or less with intensive usage (e., social sens- ing [2]). As a result, users are required to charge their devices frequently. The most common practice for users is to charge their phones by connecting them to a wall outlet through charging cables. This requires users to carry a charging cable and find an outlet, which is mostly available indoors.

Thus, the charging process can potentially be irritating and sometimes infeasible. With the integration of built-in wireless charging capability in recent phones (including iPhone 8 and X [3]), users are relieved from the need to carry charging cables but the current application of wireless charging is very limited as it requires not the phone but the charging mat to be con- nected to an outlet.