Wi Fi Economic Topics Jon Crowcroft jon crowcroftcl

  • Slides: 13
Download presentation
Wi. Fi Economic Topics Jon Crowcroft jon. crowcroft@cl. cam. ac. uk 1

Wi. Fi Economic Topics Jon Crowcroft jon. crowcroft@cl. cam. ac. uk 1

Mobileman Kickoff Meeting Pisa, 5/11/02 • Wi. Fi and p 2 p • Spectrum

Mobileman Kickoff Meeting Pisa, 5/11/02 • Wi. Fi and p 2 p • Spectrum Economics 2

Some possible 2 ndary collaborators • In CMI, we have • Richard Steinberg, JIMS,

Some possible 2 ndary collaborators • In CMI, we have • Richard Steinberg, JIMS, Cambridge – RF Auctions • David Newbury, Applied Economics, Cambridgecomponent markets, externalities etc • Frank Kelly, Stats Lab, Cambridge - congestion charging • Tolga Uzuner, Ph. D student, complexity of charging • DDC/Karen Sollins et al at MIT LCS • RCA (Robert |Leese at Oxford), Ofcomm – other regulators? – e. g. The EU 3

Market Models • Re-run M 3 I (cluster with M 3 I followup maybe,

Market Models • Re-run M 3 I (cluster with M 3 I followup maybe, see MMAPPS perhaps) • • Look at welfare/greed game models Look at regulation Look at resale, spot and futures Look at hybrid models • Auditability • Costs of Billing systems • Expectations, and settlements (esp. when Wi. Fi uses wired core!) 4

Online Auctions • • Smart Market (Varian et al) & Vickrey Auction. . Problem

Online Auctions • • Smart Market (Varian et al) & Vickrey Auction. . Problem of Value of boundary nodes Double Trouble of boundary to fixed network Triple Trouble of boundary to 2 or 3 wireless (Wi. Fi, GPRS, Fixed) 5

Congestion $ • • Loss based – hopeless! ECN based – assumes bottleneck Reality

Congestion $ • • Loss based – hopeless! ECN based – assumes bottleneck Reality is congestion is in region Need to raise price of use within region How to distinguish local and transit users? Needs network centric monitoring and control Needs edge system too, since in MANET, network control point is the end system! 6

Timescales for Congestion $ • Choices: • • Packet Based TCP connection based Session

Timescales for Congestion $ • Choices: • • Packet Based TCP connection based Session Based Subscription Based – c. f Virtual Communities… • Spatial Cost model? • Proportional Fairness model applied to MANET? • Linear with distance, or relate to Tx & Rx Battery Cost and Interference model (Gupta et al!) 7

Policing Problems • • • Prevention v Detection Refund models? Integration with existing bill/police

Policing Problems • • • Prevention v Detection Refund models? Integration with existing bill/police systems Integrate AAA security with MAC/Police? Tamper-proof modules? (c. f. M 3 I edge meter tamper-proof code!) • Risk Management (c. f. Ph. Ds in Cambridge TRUST and TAPAS IST projects|) – auditor rules on acceptable risks! 8

Virtual Communities (Wi. Fi gridlets) • Need to have a system for discovery/creation of

Virtual Communities (Wi. Fi gridlets) • Need to have a system for discovery/creation of virtual communities (all taxi drivers, all trainspotters etc) • Join/leave/invite/evict model • Need to decide how to distribute the database of communities and their wireless network requirements – CAN, central, full replication, or what? • Mixture 9

Spectrum Trade Model • Auctions • English, Dutch, Vickrey etc • Game theoretic framework

Spectrum Trade Model • Auctions • English, Dutch, Vickrey etc • Game theoretic framework applies clearly • Component systems • Fixed+wireless • Neighbour ownership • Congestion price model – dynamics etc 10

Field Theoretic Model • • • ECN++ every packet carries force field info Can

Field Theoretic Model • • • ECN++ every packet carries force field info Can be repulsion as a function of distance Measurement of flux into/out of region: Can model this to match underlying network Dynamics and static capability as well as economics • Would be nice to build this as experiment 11

Cambridge Mobileman P 2 P Research Topics • • Market Models Virtual Communities (Wi.

Cambridge Mobileman P 2 P Research Topics • • Market Models Virtual Communities (Wi. Fi. Grids? ) Allocation Auctions Field Theory metaphor to unify? 12

Non Topics for Cambridge • We don’t do • Wireless MAC • Fixing 802.

Non Topics for Cambridge • We don’t do • Wireless MAC • Fixing 802. 11 multi-hop! • RF propagation or aerial design 13