On Cooperative Wireless Techniques On Cooperative Wireless Techniques A WINLAB Research Sampling A WINLAB Research Sampling Predrag Spasojevi Spasojevi ć ć Predrag WINLAB, Rutgers University WINLAB, Rutgers University Industrial Advisory Board Meeting, December 4, 2007
Collaborate to Enable Collaborate to Enable • Communication Reliability Communication Reliability • • • Message Confidentiality Message Confidentiality • Efficient Sensing Efficient Sensing • • Efficient Data Collection Efficient Data Collection • • Efficient Spectrum Usage Efficient Spectrum Usage • • • Spectrum Usage Analysis Spectrum Usage Analysis • Interference Mitigation, and Interference Mitigation, and … … •
Cooperative Techniques for Ensuring Cooperative Techniques for Ensuring Communication Reliability and Message Confidentiality Communication Reliability and Message Confidentiality
Transmitter Cooperation for Confidentiality Transmitter Cooperation for Confidentiality W 1 Reliability W 2 Confidentiality • Trusted users can cooperate to defend adversarial eavesdropping. Xiaojun Tang, Predrag Spasojevic, Ruoheng Ruoheng Liu, and H. Vincent Poor Xiaojun Tang, Predrag Spasojevic, Liu, and H. Vincent Poor
Receiver ARQ Feedback Improves Receiver ARQ Feedback Improves Reliability and Message Confidentiality Reliability and Message Confidentiality ^ W Reliability W User 1 Transmitter Secrecy W User 2 • Simple retransmission may hurt secrecy. Can retransmission benefit secrecy? Yes! If done properly. Xiaojun Tang, Predrag Spasojevic, Ruoheng Ruoheng Liu, and H. Vincent Poor Xiaojun Tang, Predrag Spasojevic, Liu, and H. Vincent Poor
Receiver ARQ Feedback Improves Receiver ARQ Feedback Improves Reliability and Message Confidentiality Reliability and Message Confidentiality Secrecy Throughput v.s. Main SNR Secrecy Throughput v.s. Main SNR ζ Target reliability outage prob.: =0.001 e 2.5 RTD – Repetition Time Diversity 2 INR – Incremental Redundancy ζ s Throughput 1.5 INR Eavesdropper average SNR: 5dB 1 RTD ζ Target secrecy outage prob.: = 0.001 s Maximum # of transmissions: M=8 0.5 0 10 11 12 13 14 15 16 17 18 19 20 Main SNR RTD may outperform INR, when there is a reliability outage constraint Xiaojun Tang, Predrag Spasojevic, Ruoheng Ruoheng Liu, and H. Vincent Poor Xiaojun Tang, Predrag Spasojevic, Liu, and H. Vincent Poor
Collaborative Sensing and Data Collection Collaborative Sensing and Data Collection
Collaborative Radio Scene Analysis one Bluetooth and two 802.11b Nodes sensors BT node WLAN nodes Goran Ivkovic, Predrag Spasojevic, Predrag Spasojevic, and Ivan Seskar
Radio Scene Analysis: Spectrogram Reconstruction recovered before BT packets WLAN packets Goran Ivkovic, Predrag Spasojevic, Predrag Spasojevic, and Ivan Seskar
Radio Scene Analysis: Recovery of Transmitter Activity Patterns and PSDs WLAN BT Goran Ivkovic, Predrag Spasojevic, Predrag Spasojevic, and Ivan Seskar
Data Collection in Location-Unaware Networks collaborative distributed dissemination infrastructure building a circular route infrastructure model R 2 R 1 Silvija Kokalj-Filipovic, Roy Yates, Predrag Spasojevic, Predrag Spasojevic,
Coding for Data Storage and Random Access Collection data dissemination and storage: relaying, overhearing, and random coding n 1 2 3 strategies for efficient decoding random access push-pull data collection push: immediate neighborhood access pull: search for desired coded packets Silvija Kokalj-Filipovic, Predrag Spasojevic, Predrag Spasojevic, Roy Yates
Coding for Data Storage and Random Access Collection Random access push-pull data collection and decoding How many coded packets do I need to pull to decode all data? LEGEND D: Fountain with Degree-2 Doping U: uniform doping 100 D min DP D mean DP 90 D max DP U min DP U mean DP 80 U max DP 70 percentage od doping symbols 60 50 random packet pull 40 30 20 10 0 0 500 1000 1500 2000 2500 3000 3500 4000 n: number of symbols to decode “smart” packet pull Silvija Kokalj-Filipovic, Predrag Spasojevic, Predrag Spasojevic, Roy Yates
Collaborative Signal Processing in Energy Replenishable Sensor Networks Jing Lei, Roy Yates and Larry Greenstein Energy-Aware Transmission Policy Modeled by Markov Sources Available for Sensor Replenishment Chain mechanical energy � thermal energy � α : replacement radiant energy � electromagnetic energy � β β β α β + β : recharging B E 55 0 3 1 2 4 λ λ λ λ λ 5,3 5,4 5,5 5,2 5,1 D A Energy Level low high C
• Collaborative MMSE Estimation of Gaussian Markov Processes by Replenishable Sensors Jing Lei, Roy Yates and Larry Greenstein • Collaborative Detection Based on Kullback Leibler Distance by Replenishable Sensors Jing Lei, Hang Liu (Thomson CR), Roy Yates and Larry Greenstein – Each sensor calculates histogram of observations – K-L distance between observation histogram and underlying pdf is calculated – Soft-decision based on quantized K-L distance
Will Radios Collaborate? And Will Radios Collaborate? And How to Encourage Collaboration? How to Encourage Collaboration?
Coalitions in Cooperative Networks Why should rational self-interested users cooperate? – Do all users gain (greater rate) from cooperation? – Can users gain more by cooperating only selectively? – What if there are costs to cooperation? (e.g. decode other users’ signals) • Users may prefer to form coalitions. • A cooperative protocol is stable if no subset of users defects to form a coalition • Do stable forms of cooperation always exist? Suhas Mathur, Lalitha Sankara Narayanan and Narayan B. Mandayam
Coalitions in Cooperative Networks • Receiver cooperation – The grand coalition (coalition of all user) always forms – Bargaining theory can be used to guarantee fair allocations of rate to users. • Transmitter cooperation with ideal inter-user links – User are deterred fro defection by the threat of jamming interference from other users – Grand coalition is the only possible stable structure but doesn’t always form • User cooperation using partial-decode-and-forward – Even in situations where cooperation is most expected (clustered users in a MAC) the coop. of all users cannot be guaranteed – Depends on strength of inter-user links and the powers of individual users Suhas Mathur, Lalitha Sankara Narayanan and Narayan B. Mandayam
Bandwidth Exchange as an Incentive for Relaying A node delegates a portion of its bandwidth to relay in exchange for cooperation. Nash bargaining solution to efficiently and fairly explore the advantages of bandwidth exchange. Dan Zhang and Narayan B. Mandayam
To Collaborate or Not? To Collaborate or Not? for Spectrum Access and Interference Mitigation for Spectrum Access and Interference Mitigation
Dynamic Spectrum Allocation for Uplink Users with Heterogeneous Utilities Utility based spectrum allocation � Multiple users and SPs � SPs allocate spectrum to users � Spectrum Regulator (Govt. org. like FCC) User applications: utility functions of Level I � SPs share spectrum spectrum amongst themselves X i Users transmit to SPs (uplink) � SPs have different efficiencies � Service Providers (SP) (Base Stations) User sum utility maximization � x ij p ij r ij Level II Distributed implementation SPs provide � spectrum to end users SPs charge users spectrum price � End Users Users maximize utility minus spectrum � costs Joydeep Acharya, Roy D. Yates
Dynamic Spectrum Allocation for Uplink Users with Heterogeneous Utilities Optimal Spectrum Allocation � At optimal price all spectrum is used � A user obtains spectrum from only one SP � Each user is allocated spectrum � User with higher marginal utility gets more � User with better link gain/power gets more � More users imply higher spectrum price Joydeep Acharya, Roy D. Yates
Coordinated vs Distributed Access Scheduling for Variable Rate Links 1 4 1 2 3 3 Spectrum server devises a centralized time schedule for the links to : • – maximize sum rate of the network – implement fair scheduling – implement energy efficient scheduling – meet a given set of rate requirements on the links • Centralized scheduling provide upper bounds to system performance Chandrasekharan Raman, R. Yates, N. Mandayam
Coordinated vs Distributed Access Scheduling for Variable Rate Links • Each link transmits with a chosen probability (independent of other links) • Average link rates depends on the interference from the other link • Provide a decentralized algorithm to achieve a desired rate vector r 2 r 2 r 1 r 1 High interference case = Rates for low interference case = same as rates strictly smaller than rate in in centralized scheme centralized scheme Chandrasekharan Raman, Jasvinder Singh, R. Yates, N. Mandayam
Cooperative Base Station Transmissions with Multiple Antennas Coordinate the base antenna transmissions so as to minimize the inter-cell interference, and hence to increase the downlink system capacity. M. Kemal Karakayali, J. Foschini, R. Valenzuela, Roy Yates
Recommend
More recommend
