1. Wireless Communications Overview 1

2. 2 Wireless History  Ancient Systems: Smoke Signals, Carrier Pigeons, … Radio invented in the 1880s by Marconi  Many sophisticated military radio systems were developed during and after WW2  Cellular has enjoyed exponential growth since 1988, with almost 5 billion users worldwide today Ignited the wireless revolution Voice, data, and multimedia becoming ubiquitous Use in third world countries growing rapidly  Wifi also enjoying tremendous success and growth Wide area networks (e.g. Wimax) and short-range systems other than Bluetooth (e.g. UWB) less successful

3. 3 Future Wireless Networks Ubiquitous Communication Among People and Devices Next-generation Cellular Wireless Internet Access Wireless Multimedia Sensor Networks Smart Homes/Spaces Automated Highways In-Body Networks All this and more …

4. 4 Challenges Cellular Mem BT CPU GPS WiFi mmW Cog Radio Network/Radio Challenges Gbps data rates with “no” errors Energy efficiency Scarce/bifurcated spectrum Reliability and coverage Heterogeneous networks Seamless internetwork handoff AdHoc Short-Range 5G5G Device/SoC Challenges Performance Complexity Size, Power, Cost High frequencies/mmWave Multiple Antennas Multiradio Integration Coexistance

5. Software-Defined (SD) Radio: Multiband antennas and wideband A/Ds span BW of desired signals DSP programmed to process desired signal based on carrier frequency, signal shape, etc. Avoid specialized HW Cellular Apps Processor BT Media Processor GPS WLAN Wimax DVB-H FM/XM A/D DSP A/D Is this the solution to the device challenges? Today, this is not cost, size, or power efficient Compressed sensing may be a solution for sparse signals

6. Careful what you wish for… 6 Exponential Mobile Data Growth Leading to massive spectrum deficit Growth in mobile data, massive spectrum deficit and stagnant revenues require technical and political breakthroughs for ongoing success of cellular Source: Unstrung Pyramid Research 2010 Source: FCC

7. Scarce Wireless Spectrum and Expensive $$$ 7

8. On the Horizon: “The Internet of Things” Number of Connected Objects Expected to Reach 50bn by 2020

9. “Effectively unlimited” capacity possible via personal cells (pcells). S. Perlman, Artemis. “The wireless industry has reached the theoretical limit of how fast networks can go” K. Fitcher, Connected Planet “There is no theoretical maximum to the amount of data that can be carried by a radio channel” M. Gass, 802.11 Wireless Networks: The Definitive Guide “We’re 99% of the way” to the “barrier known as Shannon’s limit,” D. Warren, GSM Association Sr. Dir. of Tech. Shannon was wrong, there is no limit We are at the Shannon Limit Are we at the Shannon limit of the Physical Layer?

10. What would Shannon say? Time-varying channels. We don’t know the Shannon capacity of most wireless channels Channels with interference or relays. Cellular systems Channels with delay/energy/$$$ constraints. Ad-hoc and sensor networks Shannon theory provides design insights and system performance upper bounds

11. Current/Emerging Wireless Systems Current: 4G Cellular Systems (LTE-Advanced) 4G Wireless LANs/WiFi (802.11ac) Satellite Systems Bluetooth Zigbee WiGig Emerging 5G Cellular and WiFi Systems mmWave Systems Ad/hoc and Cognitive Radio Networks Energy-Harvesting Systems Much room For innovation

12. Wireless networks are everywhere, yet… - Connectivity is fragmented - Capacity is limited (spectrum crunch and interference) - Roaming between networks is ad hoc TV White Space & Cognitive Radio

13. Spectral Reuse Due to its scarcity, spectrum is reused BS In licensed bands Cellular, Wimax Wifi, BT, UWB,… and unlicensed bands Reuse introduces interference 13

14. Cellular Systems: Reuse channels to maximize capacity  Geographic region divided into cells  Frequency/timeslots/codes/ reused at spatially-separated locations.  Co-channel interference between same color cells.  Base stations/MTSOs coordinate handoff and control functions  Shrinking cell size increases capacity, as well as networking burden BASE STATION MTSO 14

15. Cellular Phones Everything Wireless in One Device 15

16. Cellular Phones Much better performance and reliability than today - Gbps rates, low latency, 99% coverage indoors and out BS Phone System BS San Francisco Paris N th -Gen Cellular N th -Gen Cellular Internet LTE backbone is the Internet Everything wireless in one device Burden for this performance is on the backbone network

17. 4G/LTE Cellular  Much higher data rates than 3G (50-100 Mbps) 3G systems has 384 Kbps peak rates  Greater spectral efficiency (bits/s/Hz) Through MIMO, adaptive techniques, “ICIC”  Flexible use of up to 100 MHz of spectrum 20 MHz spectrum allocation common  Low packet latency (<5ms).  Reduced cost-per-bit  Support for multimedia  All IP network

18. Rethinking “Cells” in Cellular  Traditional cellular design “interference-limited” MIMO/multiuser detection can remove interference Cooperating BSs form a MIMO array: what is a cell? Relays change cell shape and boundaries Distributed antennas move BS towards cell boundary Small cells create a cell within a cell Mobile cooperation via relays, virtual MIMO, network coding. Small Cell Relay DAS Coop MIMO How should cellular systems be designed? Will gains in practice be big or incremental; in capacity or coverage?

19. Are small cells the solution to increase cellular system capacity? Yes, with reuse one and adaptive techniques (Alouini/Goldsmith 1999) A=  D 2  Area Spectral Efficiency S/I increases with reuse distance (increases link capacity). Tradeoff between reuse distance and link spectral efficiency (bps/Hz). Area Spectral Efficiency: A e =  R i /(  D 2  ) bps/Hz/Km 2.

20. The Future Cellular Network: Hierarchical Architecture Future systems require Self-Organization (SON) and WiFi Offload 10x Lower COST/Mbps 10x CAPACITY Improvement Near 100% COVERAGE (more with WiFi Offload) MACRO: solving initial coverage issue, existing network PICO: solving street, enterprise & home coverage/capacit y issue Macrocell Picocell Femtocell Today’s architecture 3M Macrocells serving 5 billion users Anticipated 1M small cells per year

21. SON Premise and Architecture Node Installation Initial Measurements Self Optimization Self Healing Self Configurati on Measureme nt SON Server 21 SoN Server Macrocell BS Mobile Gateway Or Cloud Small cell BS X2 IP Network SW Agent  SON is part of 3GPP/LTE standard

22. Green” Cellular Networks  Minimize energy at both the mobile and base station via New Infrastuctures: cell size, BS placement, DAS, Picos, relays New Protocols: Cell Zooming, Coop MIMO, RRM, Scheduling, Sleeping, Relaying Low-Power (Green) Radios: Radio Architectures, Modulation, coding, MIMO  Minimize energy at both the mobile and base station via New Infrastuctures: cell size, BS placement, DAS, Picos, relays New Protocols: Cell Zooming, Coop MIMO, RRM, Scheduling, Sleeping, Relaying Low-Power (Green) Radios: Radio Architectures, Modulation, coding, MIMO Pico/Femto Relay DAS Coop MIMO How should cellular systems be redesigned for minimum energy? Research indicates that significant savings is possible

23. Why Green, why now?  The energy consumption of cellular networks is growing rapidly with increasing data rates and numbers of users  Operators are experiencing increasing and volatile costs of energy to run their networks  There is a push for “green” innovation in most sectors of information and communication technology (ICT)  There is a wave of consortia and government programs focused on green wireless

24. Wifi Networks Multimedia Everywhere, Without Wires 24 802.11ac Wireless HDTV and Gaming Streaming video Gbps data rates High reliability Coverage inside and out Wifi is today’s small cell

25. Wireless Local Area Networks (WLANs) WLANs connect “local” computers (100m range) Breaks data into packets Channel access is shared (random access + backoff) Backbone Internet provides best-effort service Poor performance in some apps (e.g. video) 01011011 Internet Access Point 0101 1011 25

26. 26 Evolution of WLAN 1999 Next 10 years 2.4GHz ， 20 MHz > 54 Mbps 2.4/5GHz, 20 - 40MHz 600 Mbps IMT-Advanced 20-40 MHz ==> 1 Gbps 2003 2009 2012 2.4/5GHz, 20MHz 11/54 Mbps Single Link from MAC layer to upper layer > 500 Mbps; Multi-user communication >1 Gbps. 802.11ac is based on 802.11n enhanced ： OFDM/OFDMA 、 MIMO 、 SDMA 、 FEC 、 MAC layer optimization 、 Interference Coordination 、 Cognitive Radio and Carrier Aggregation

27. Wireless LAN Standards  802.11b (Old – 1990s) Standard for 2.4GHz ISM band (80 MHz) Direct sequence spread spectrum (DSSS) Speeds of 11 Mbps, approx. 500 ft range  802.11a/g (Middle Age– mid-late 1990s) Standard for 5GHz band (300 MHz)/also 2.4GHz OFDM in 20 MHz with adaptive rate/codes Speeds of 54 Mbps, approx. 100-200 ft range  802.11n/ac(current) Standard in 2.4 GHz and 5 GHz band Adaptive OFDM /MIMO in 20/40/80/160 MHz Antennas: 2-4, up to 8 Speeds up to 600Mbps / 10 Gbps), approx. 200 ft range Other advances in packetization, antenna use, multiuser MIMO Many WLAN cards have all 3 (a/b/g/n)

28. Why does WiFi performance suck?  The WiFi standard lacks good mechanisms to mitigate interference, especially in dense AP deployments Multiple access protocol (CSMA/CD) from 1970s Static channel assignment, power levels, and carrier sensing thresholds In such deployments WiFi systems exhibit poor spectrum reuse and significant contention among APs and clients Result is low throughput and a poor user experience Carrier Sense Multiple Access: if another WiFi signal detected, random backoff Collision Detection: if collision detected, resend

29. Why not use SoN for WiFi? SoN-for-WiFi: dynamic self-organization network software to manage of WiFi APs. Allows for capacity/coverage/interference mitigation tradeoffs. Also provides network analytics and planning. SoN Controller - Channel Selection - Power Control - etc.

30. In fact, why not use SoN for all wireless networks? Software-Defined Wireless Networks TV White Space & Cognitive Radio mmWave networks Vehicle networks

31. Software-Defined Network Virtualization Network virtualization combines hardware and software network resources and functionality into a single, software-based virtual network Wireless Network Virtualization Layer

32. Software-Defined Wireless Network (SDWN) Architecture WiFi Cellular mmWave Cognitive Radio Freq. Allocation Freq. Allocation Power Control Power Control Self Healing Self Healing ICIC QoS Opt. QoS Opt. CS Threshold CS Threshold UNIFIED CONTROL PLANE HW Layer SW layer App layer Video Security Vehicular Networks Vehicular Networks Health M2M

33. WiGig and mmWave  WiGig Standard operating in 60 GHz band Data rates of 7-25 Gbps Bandwidth of around 10 GHz (unregulated) Range of around 10m (can be extended) Uses/extends 802.11 MAC Layer Applications include PC peripherals and displays for HDTVs, monitors & projectors  mmWave Couples 60GHz with massive MIMO and better MAC Promises long-range communication w/Gbps data rates Hardware, propagation and system design challenges Much research on this topic today

34. Wimax (802.16)  Wide area wireless network standard System architecture similar to cellular Hopes to compete with cellular  OFDM/MIMO is core link technology  Operates in 2.5 and 3.5 MHz bands Different for different countries, 5.8 also used. Bandwidth is 3.5-10 MHz  Fixed (802.16d) vs. Mobile (802.16e) Wimax Fixed: 75 Mbps max, up to 50 mile cell radius Mobile: 15 Mbps max, up to 1-2 mile cell radius 34

35. Satellite Systems  Cover very large areas  Different orbit heights GEOs (39000 Km) versus LEOs (2000 Km) http://www.globalcomsatphone.com/globalcom/leo_geo.html  Optimized for one-way transmission Radio (XM, Sirius) and movie (SatTV, DVB/S) broadcasts Most two-way systems struggling or bankrupt  Global Positioning System (GPS) use growing Satellite signals used to pinpoint location Popular in cell phones, PDAs, and navigation devices 35

36. Paging Systems  Broad coverage for short messaging  Message broadcast from all base stations  Simple terminals  Optimized for 1-way transmission  Answer-back hard  Overtaken by cellular 36

37. 8C32810.61-Cimini-7/98 Bluetooth  Cable replacement RF technology (low cost)  Short range (10m, extendable to 100m)  2.4 GHz band (crowded)  1 Data (700 Kbps) and 3 voice channels, up to 3 Mbps  Widely supported by telecommunications, PC, and consumer electronics companies  Few applications beyond cable replacement  Bluetooth 4.0, June 30, 2010 Classic Bluetooth Bluetooth high speed (based on WiFi) Bluetooth low energy  http://en.wikipedia.org/wiki/Bluetooth http://en.wikipedia.org/wiki/Bluetooth 37

38. 38 Ultrawideband Radio (UWB)  UWB is an impulse radio: sends pulses of tens of picoseconds(10-12) to nanoseconds (10-9) Duty cycle of only a fraction of a percent  A carrier is not necessarily needed  Uses a lot of bandwidth (GHz)  High data rates, up to 500 Mbps  7.5 Ghz of “free spectrum” in the U.S. (underlay)  Multipath highly resolvable: good and bad  Limited commercial success to date

39. IEEE 802.15.4 / ZigBee Radios  Low-Rate WPAN  Data rates of 20, 40, 250 Kbps  Support for large mesh networking or star clusters  Support for low latency devices  CSMA-CA channel access  Very low power consumption  Frequency of operation in ISM bands Focus is primarily on low power sensor networks 39

40. Tradeoffs ZigBee Bluetooth 802.11b 802.11g/a 3G UWB Range Rate Power 802.11n 40

41. Spectrum Regulation  Spectrum a scarce public resource, hence allocated  Spectral Allocation in US controlled by FCC (commercial) or OSM (defense)  FCC auctions spectral blocks for set applications.  Some spectrum set aside for universal use  Worldwide spectrum controlled by ITU-R  Regulation is a necessary evil. in multiple cognitive radio Innovations in regulation being considered worldwide paradigms 41

42. Standards  Interacting systems require standardization  Companies want their systems adopted as standard Alternatively try for de-facto standards  Standards determined by TIA/CTIA in US IEEE standards often adopted Process fraught with inefficiencies and conflicts  Worldwide standards determined by ITU-T In Europe, ETSI is equivalent of IEEE

43. Emerging Systems * Software-defined and low-complexity radios Cognitive radio networks Ad hoc/mesh wireless networks Sensor networks Distributed control networks The smart grid Biomedical networks

44. Compressed Sensing  Basic premise is that signals with some sparse structure can be sampled below their Nyquist rate  Signal can be perfectly reconstructed from these samples by exploiting signal sparsity This significantly reduces the burden on the front-end A/D converter, as well as the DSP and storage  Might be key enabler for SD and low-energy radios Only for incoming signals “sparse” in time, freq., space, etc.

45. Reduced-Dimension Communication System Design Compressed sensing ideas have found widespread application in signal processing and other areas. Basic premise of CS: exploit sparsity to approximate a high-dimensional system/signal in a few dimensions. Can sparsity be exploited to reduce the complexity of communication system design?

46. Sparsity: where art thou? Sparse state space: e.g reduced-dimension network control Sparse users: e.g. reduced-dimension multiuser detection To exploit sparsity, we need to find communication systems where it exists Sparse samples: e.g. sub-Nyquist sampling Sparse signals: e.g. white-space detection

47. Sparse Samples For a given sampling mechanism (i.e. a “new” channel) What is the optimal input signal? What is the tradeoff between capacity and sampling rate? What known sampling methods lead to highest capacity? What is the optimal sampling mechanism? Among all possible (known and unknown) sampling schemes Sampling Mechanism (rate f s ) New Channel

48. Capacity under Sub-Nyquist Sampling Theorem 1: Theorem 2: Bank of Modulator+Filter  Single Branch  Filter Bank Theorem 3: Optimal among all time-preserving nonuniform sampling techniques of rate f s (ISIT’12; Arxiv) zzzzzz zzzz equals zzzz zzzz zz

49.  Selects the m branches with m highest SNR Example (Bank of 2 branches) Joint Optimization of Input and Filter Bank highest SNR 2 nd highest SNR low SNR low SNR Capacity monotonic in f s How does this translate to practical modulation and coding schemes

50. Reduced-Dimension Network Design Random State Evolution Sampling and Inference Stochastic Control Resource Management Reduced-Dimension Representation

51. Cognitive Radios  Cognitive radios support new users in existing crowded spectrum without degrading licensed users Utilize advanced communication and DSP techniques Coupled with novel spectrum allocation policies  Multiple paradigms (MIMO) Underlay (interference below a threshold) Interweave finds/uses unused time/freq/space slots Overlay (overhears/relays primary message while cancelling interference it causes to cognitive receiver) NCR IPIP CR CRRx NCRRx NCRTx CRTx MIMO Cognitive Underlay Cognitive Overlay

52. Cognitive Radio Multiple Paradigms  Underlay Cognitive radios constrained to cause minimal interference to noncognitive radios  Interweave Cognitive radios find and exploit spectral holes to avoid interfering with noncognitive radios  Overlay Cognitive radios overhear and enhance noncognitive radio transmissions Knowledge and Complexity 52

53. Underlay Systems  Cognitive radios determine the interference their transmission causes to noncognitive nodes Transmit if interference below a given threshold  The interference constraint may be met Via wideband signalling to maintain interference below the noise floor (spread spectrum or UWB) Via multiple antennas and beamforming NCR IPIP CR

54. Interweave Systems  Measurements indicate that even crowded spectrum is not used across all time, space, and frequencies Original motivation for “cognitive” radios (Mitola’00)  These holes can be used for communication Interweave CRs periodically monitor spectrum for holes Hole location must be agreed upon between TX and RX Hole is then used for opportunistic communication with minimal interference to noncognitive users

55. Overlay Systems  Cognitive user has knowledge of other user’s message and/or encoding strategy Used to help noncognitive transmission Used to presubtract noncognitive interference RX1 RX2 NCR CR

56. Performance Gains from Cognitive Encoding Only the CR transmits outer bound our scheme prior schemes

57. ce Ad-Hoc/Mesh Networks Outdoor Mesh Indoor Mesh 57

58. Ad-Hoc Networks  Peer-to-peer communications No backbone infrastructure or centralized control  Routing can be multihop.  Topology is dynamic.  Fully connected with different link SINRs  Open questions Fundamental capacity region Resource allocation (power, rate, spectrum, etc.) Routing

59. Cooperation in Wireless Networks  Many possible cooperation strategies: Virtual MIMO, relaying (DF, CF, AF), one-shot/iterative conferencing, and network coding Nodes can use orthogonal or non-orthogonal channels. Many practice and theoretical challenges New full duplex relays can be exploited

60. Design Issues  Ad-hoc networks provide a flexible network infrastructure for many emerging applications.  The capacity of such networks is generally unknown.  Transmission, access, and routing strategies for ad- hoc networks are generally ad-hoc.  Crosslayer design critical and very challenging.  Energy constraints impose interesting design tradeoffs for communication and networking. 60

61. General Relay Strategies Can forward message and/or interference Relay can forward all or part of the messages Much room for innovation Relay can forward interference To help subtract it out TX1 TX2 relay RX2 RX1 X1X1 X2X2 Y 3 =X 1 +X 2 +Z 3 Y 4 =X 1 +X 2 +X 3 +Z 4 Y 5 =X 1 +X 2 +X 3 +Z 5 X 3 = f(Y 3 )

62. Beneficial to forward both interference and message For large powers, this strategy approaches capacity

63. 63 Wireless Sensor Networks Data Collection and Distributed Control  Energy (transmit and processing) is the driving constraint  Data flows to centralized location (joint compression)  Low per-node rates but tens to thousands of nodes  Intelligence is in the network rather than in the devices Smart homes/buildings Smart structures Search and rescue Homeland security Event detection Battlefield surveillance

64. Energy-Constrained Nodes  Each node can only send a finite number of bits. Transmit energy minimized by maximizing bit time Circuit energy consumption increases with bit time Introduces a delay versus energy tradeoff for each bit  Short-range networks must consider transmit, circuit, and processing energy. Sophisticated techniques not necessarily energy-efficient. Sleep modes save energy but complicate networking.  Changes everything about the network design: Bit allocation must be optimized across all protocols. Delay vs. throughput vs. node/network lifetime tradeoffs. Optimization of node cooperation. 64

65. Channel Coding  Channel coding used to reduce error probability Metric is typically the coding gain Also measured against Shannon limit  Block and Convolutional Codes Induce bandwidth expansion, reduce data rate Easy to implement; well understood (decades of research)  Coded Modulation (Trellis/Lattice Codes; BICM) Joint design of code and modulation mapper Provides coding gain without bandwidth expansion Can be directly applied to adaptive modulation  Turbo Codes (within a fraction of a dB of Shannon limit) Clever encoding produces pseudo “random” codes easily Decoder fairly complex  Low-density parity check (LDPC) Codes State-of-the-art codes (802.11n, LTE) Invented by Gallager in 1950s, reinvented in 1990s Low density parity check matrix Encoder complex; decoder uses belief propagation techniques Coded Uncoded Coding Gain SNR PbPb Shannon Limit

66. Green Codes (for short distances) Is Shannon-capacity still a good metric for system design?

67. Coding for minimum total power Is Shannon-capacity still a good metric for system design? Extends early work of El Gamal et. al.’84 and Thompson’80 Computational Nodes On-chip interconnects

68. Fundamental area-time-performance tradeoffs  For encoding/decoding “good” codes,  Stay away from capacity! Close to capacity we have Large chip-area, more time/power Area occupied by wires Encoding/decoding clock cycles Regular LDPCs closer to bound than capacity-approaching LDPCs! Need novel code designs with short wires, good performance

69. 69 Distributed Control over Wireless Automated Vehicles - Cars - Airplanes/UAVs - Insect flyers Interdisciplinary design approach Control requires fast, accurate, and reliable feedback. Wireless networks introduce delay and loss Need reliable networks and robust controllers Mostly open problems : Many design challenges

70. The Smart Grid: Fusion of Sensing, Control, Communications carbonmetrics.eu Open problems: - Optimize sensor placement in the grid - New designs for energy routing and distribution - Develop control strategies to robustify the grid - Look at electric cars for storage and path planning

71. Much progress in some areas  Smart metering  Green buildings and structures  Optimization  Demand response  Modeling and simulation  Incentives and economics But many of the hard research questions have not yet been asked Current work: sensor placement, fault Detection and control with sparse sensors

72. Applications in Health, Biomedicine and Neuroscience Recovery from Nerve Damage Doctor-on-a-chip Wireless Network Neuro/Bioscience - EKG signal reception/modeling - Brain information theory - Nerve network (re)configuration - Implants to monitor/generate signals -In-brain sensor networks - SP/Comm applied to bioscience Body-Area Networks

73. scan Gene Expression Profiling tumor tissue RNA extraction labeling hybridization 70 genes  Gene expression profiling predicts clinical outcome of breast cancer (Van’t Veer et al., Nature 2002.) Immune cell infiltration into tumors  good prognosis.  Gene expression measurements: a mix of many cell types Gene Signatures Cell-type Proportion Cell Types

74. Looks like CDMA “despreading” Many gene expression deconvolution algorithms exist  Shen-Orr et al., Nature methods 2010  known “C” and “k”  Lu et al., PNAS 2003  known “G” and “k”  Vennet et al., Bioinformatics 2001  known “k”  Large databases exist where these parameters are unknown Can we apply signal processing methods to blindly separate gene expression? We adapt techniques from hyperspectroscopy (Piper et al, AMOS 2004) assuming “C”, “G” and “k” unknown Beat existing techniques, even nonblind ones

75. BA CD E Direct information (DI) inference Constrained DI inference BA CD E Neuron layout BA CD E DI inference with delay lower bound BA CD E Pathways through the brain

76. Other Applications of Communications and Signal Processing to Brain Science  Epilepsy Epileptic fits caused by an oscillating signal moving from one region to another. When enough regions oscillate, a fit occurs Working with epilepsy expert (MD) to understand how signal travels between regions. Has sensors directly implanted in brain: can read signals and inject them. Can we use drugs to block propagation or signal injection to cancel signals  Parkinsons Creates 20 MHz noise in brain region 120 MHz square wave injection mitigates the symptoms

77. Main Points  The wireless vision encompasses many exciting applications  Technical challenges transcend all system design layers  5G networks must support higher performance for some users and extreme energy efficiency for others  Cloud-based software to dynamically control and optimize wireless networks needed (SDWN)  Innovative wireless design needed for 5G cellular/WiFi, mmWave systems, massive MIMO, and IoT connectivity  Standards and spectral allocation heavily impact the evolution of wireless technology 77