Transcript - ASU Digital Repository

New Multi-nodal Wireless Communication
System Method
Frank James
April 14, 2014
Agenda




Introduction
 Problem Domain Space
 Wireless Communication
 Wired Technologies/Wireless Technologies
Mobile & Base Station Communication
 Wireless Communication Method
 Device & Network initiated roaming
Roaming Considerations
 Power Spectral Density Estimation
 Exemplar Network Processing (Advanced)
 Cognitive Radio
 Noise “Temperature”
Hardware Design (Advanced)
 Latest Network Processing HW
 Very Large Scale Integration (VLSI)
 DFM/DFT (1st pass Si success)
2
Problem Overview

Problem Statement
 WiFi link layer does not support roaming efficiently across multiple access points
on the network side.
 WiFi and Cellular base stations are not very adaptive to the RF environment.

Adaptation to RF environment left to wireless devices.
 WiFi and Cellular technologies continue to converge.

“Smart” phones support both WiFi and Cellular communication

Link layer communication between the two technologies have not converged
in the market space [yet]

Cell sizes continue to shrink to support larger capacities
 Performance at the Network “Edge” continues to increase

Roaming needs to be high performance between [Wi-Fi & Cellular] Access
Points

WiFi roaming at link layer is not well distributed or mature


Reduce number of frames dropped during a roaming event
How to implement

Not impact existing standards (.e.g. 802.11 & UMTS) in the market place

Provide ubiquitous wireless coverage across both technologies
3
Problem Overview

Goal
Develop an advanced multi-nodal communication system that will provide an
integrated link layer for BOTH WiFi and Cellular protocols within the same cell
coverage areas.
 Build in RF Adaptation techniques into multi-nodal wireless communication system
to provide better RF service capabilities to existing infrastructure
Approach
 Leverage existing WiFi ,Cellular and wired network protocol standards
 Leverage spatial diversity to extend range or seamlessly “roam” across different
radio cell sizes
 Centralize WiFi link layer communication

Utilize existing network protocol standards to utilize banck-end unreliable
communication channels

Integrate WiFi data path with Cellular data path subsystem

Distribute WiFi link layer communication with backend system
 Create “Smart” WiFi & Cellular Base Stations

Communicate RF parametric information to backend system so better network
decisions can be made for individual devices

Improve Power Spectral Density estimates so accurate information is
communicated to backend system
 Build in RF Adaptation Control on Cell Controller


4
Communication Techniques
Approach
Leverage
existing WiFi ,Cellular
and wired network protocol
standards
Leverage spatial diversity to
extend range or seamlessly
“roam” across different radio cell
sizes
Centralize WiFi link layer
communication
5
Communication Techniques
Link Layer
Mapping to Architecture
1
3
6
Communication Techniques
UMTS Link Layer
3G UMTS

UL: SF=Separation of data ,
voice, & control channels,
SC=Separation of users
 DL: SF=Per User Separation,
SC=Unique per cell
 Many different data and control
message variants with the
general format below
Header
Slot#0
Payload: Data, SMS/Voice, or Control Information
…
Slot#14
10ms
7
Communication Techniques
WiFi Distributed Link Layer
•
•
•
•
Flow Control (FC) is “piggy
backed” with data link layer
messages so fewer frames
are required to support flow
control
Wireless Functions (FUNC)
are used to perform wireless
functions that measure the
RF environment around each
base station
802.11 & Cellular MAC are
compliant link layer
communication messages
that performed in real time
Template messages are real
time messages that are
transferred to base stations
to be used to handle real-time
message responses to
wireless client devices.
8
Communication Techniques
Integration of WiFi
Cellular Subsystems
•
•
The WiFi subsystem is
integrated with the GGSN/PGW
cellular subsystem.
The GGSN/PGW cellular
subsystem is responsible for
providing the IP gateway
services for the data services
on cellular devices to the
network
•
•
In other words this services
provides the IP NAT services
to private IP addresses
provided by the subsystem.
The WiFi subsystem is also
integrated with the cellular
authentication services hosted
through the MAP services
provided through individual
services provider network
9
Wireless Hardware
Data Path (Cellular and
WiFi)
•
•
•
Overview of the hardware
state machines in a prototype
hardware system architecture
Hardware state machines
fully autonomous termination
of wireless data frames as
well as real-time RF
management messages.
Hardware state machines
provide full flow control for
link layer communication
path
10
Communication Techniques
HW Finite State Machines
Network
Buffer
Mgmt
1…N
802.11/Cellular
Roaming
HW
Queue
Mem
CTL
Input
Fifo
Output Fifo
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
Internal
DMA
System Memory
Core
Core
Core
L2
L2
Cache
I-Cache
L2Cache
Cache I-Cache
I-Cache
HW
Queue
Frame
Transmit/
Receive
Parse,
Classify
Wireless Link Layer
802.11/Cellular
Measure
D-Cache
L3
Cache
D-Cache
L3Cache
Cache
D-Cache L3
Coherency
1
802.11 MGMT Subsys
Cellular MGMT
Subsys
802.11 and Cellular Subsystems
Handle Non-Real Time Link Layer Communication
Execute in a Symmetric Multi-Processing Configuration
1…N
Frame Buffers
1…N
IV Info
Key Info
Eth/VLAN Template
802.11 MAC Template
Cellular MAC Template
Device State
Frame Queue Addr Lst
Cryptography
Per
802.11 or Cellular
User Device
& FSM
HW Cache
Structure
2
11
Hardware Data Packet
Walkthrough (Link
Layer)
•
•
•
•
Network frame (directed toward
a wireless device is received) Either by dest. IP address or
dest. MAC address.
The network frame is moved to
system memory for processing
by downstream hardware state
machines
The frame is distributed [via the
HW Queue] toward the Wireless
MAC engine
Each hardware state machine is
interacts with the Coherency
Fabric protocol to ensure
memory caches are invalidated
and updated accordingly
Wireless MAC Engine
12
Hardware Data Packet
Walkthrough (Link
Layer)
•
•
•
•
Wireless MAC Engine receives
the address of an input frame
from the network that is directed
towards a specific frame queue
ID and input/output policy (e.g.
IP or MAC address of wireless
device)
Frame templates and a small
set of CISC like instructions that
are specialized for wireless
frame processing.
Each hardware state machine is
interacts with the Coherency
Fabric protocol to ensure
memory caches are invalidated
and updated accordingly
Output sent to HW queue that
then sends the offset to the
portion of the frame that must
be encrypted
Cryptography
13
Hardware Data Packet
Walkthrough (Link
Layer)
•
•
•
Frame is encrypted (started at
an address offset as specified
by the input policy)
Frame is sent to the HW Queue
(only address of input & output
are transmitted between
hardware state machines)
Each hardware state machine is
interacts with the Coherency
Fabric protocol to ensure
memory caches are invalidated
and updated accordingly
14
Communication Techniques
Cryptography
5
4
RNG
HW Finite State Machines
Input Fifo
Output Fifo
HW
Queue
Frame
Transmit/
Receive
Parse,
Classify
Mem
CTL
HW
Queue
2a
Mem. Cache
CryptoEngine
Mem. Cache
CryptoEngine
Mem. Cache
CryptoEngine
Mem. Cache
CryptoEngine
Mem. Cache
CryptoEngine
Internal
DMA
3
System Memory
Network
Buffer
Mgmt
1
Frame Buffer
Coherency
1…N
ADDR
ADDR
ADDR
ID
ID
ID
Key
Key
Info
KeyInfo
Info
IV
Info
IV
Info
New IV
Mem.
Cache
Mem.
Cache
Prev
IV
Instructions
Instructions
Instructions
2b
CryptoEngine
If (New IV) == (Prev IV) {
Return Error;
} else {
Store New IV
Per
IV);
FlowAddr(Prev
& FSM
} HW Cache
Structure
Continue Processing…
6
15
Communication Techniques
Random Number Generator (RNG)
•
•
•
•
Required to support many security protocol requirements (e.g.
random IV, random seeds, & random content)
• IPSEC, SSL/TLS, and DTLS are examples of protocols that
require random content
Hardware system is required to check for generation of duplicate,
consecutive random numbers per FIPS 140-2
In order to accommodate processing wireless data frames that
contain random content in IV portions of the wireless frame, the
cryptographic unit must generate at a rate of performance necessary
to satisfy hardware frame processing.
The RNG design uses asynchronous design techniques to design a
RNG subsystem, one must use the instability inherent in relying
upon propagation delays in a logic design that vary with
temperature, voltage level fluctuation, substrate differences, and
register settling time that occur through combinatorial logic designs.
This technique is used to ensure a sufficient level of Entropy is
achieved.
16
Communication Techniques
References to be Added to Thesis
Cover, T. (1972). Admissibility properties of Gilbert’s encoding for unknown source probabilities. IEEE
Transactions on Information Theory, 18:216–217.
Krichevsky, R. (1998). Laplace’s law of succession and universal encoding. IEEE Transactions on
Information
Theory, 44:296–303.
McDiarmid, C. (1989). On the method of bounded differences. In Surveys in Combinatorics, pages 148–188.
Cambridge University Press.
Miller, G. (1955). Note on the bias of information estimates. In Information theory in psychology II-B, pages
95–100.
Nemenman, I., Shafee, F., and Bialek, W. (2002). Entropy and inference, revisited. NIPS, 14.
Paninski, L. (2003). Estimation of entropy and mutual information. Neural Computation, 15:1191–1253.
Paninski, L. (2004). Estimating entropy on m bins given fewer than m samples. IEEE Transactions on
Information Theory, 50:2200–2203.
Paninski, L. (2005). Variational minimax estimation of discrete distributions under KL loss. Advances in
Neural
Information Processing Systems, 17.
Paninski, L. (2008). A coincidence-based test for uniformity given very sparsely-sampled discrete data. IEEE
Transactions on Information Theory, 54:4750–4755.
Paninski, L. and Yajima, M. (2008). Undersmoothed kernel entropy estimators. IEEE Transactions on
Information Theory, 54:4384–4388.
17
Communication Techniques
Exemplar WiFi RF Measurement Technique
•
Two 802.11 radios observe
the transmission of the Probe
Request Messages
•
•
•
•
Observed by both base
stations A & B
Though base station B
responds to the Probe
Request message, base
station A is aware of the
WiFi client
Based upon Eb/No measured
empirically from base
stations A & B information
is sent to the WiFi policy
engine on the centralized
cell controller to determine
base stations A should start
responding to base station
management and data
frames.
This is referred to as soft
“handover”
18
Communication Techniques
WiFi Device State Machine
•
•
•
•
•
•
Note: The WiFi state machine starts with
initial power up.
Most prevalent is the initiation of the
association using open authentication
(e.g. no authentication initiated)
Association and Authentication is
typically followed by an EAPOL
sequence to authenticate (typically
using PSK = Pre-Shared Key)
Traffic Indication Map (TIM) = 2008 bits;
therefore, the maximum per base station
BSS is 2008 individual devices
Note: When a WiFi device calculates a
stronger PSD it initiates a Reassociation
request to that new device
If the cell controller starts transmitting
data message and management
message responses from a closer radio,
no Reassociation process is required.
19
Communication Techniques
Cellular Device State Machine
•
•
•
Just as the case with WiFi
devices, the state machine
starts with the power up.
Paging is a cellular request
initiated from a network side
and is valid for 2G,3G, & 4G
protocol stacks.
3G is complicated with the use
of scrambling codes and per
device spreading factors/codes.
•
•
Initiated
From
Network
However, if two radios transmit
the same scrambling codes
and spreading factors are
used across multiple radios
However, in 4G a device can
use the UMTS method (from the
network side) to force a
handover
20
Communication Techniques
Cellular Device State Machine
•
•
•
•
2G/3G devices use the following process during the cell
reselection process.
Note the calculation of C1 and C2 by mobile devices that
are used to report measurements are used by base stations
to manage handovers.
In addition, neighbor lists are used to determine adjacent
cell information (e.g. frequency information, ARFCN)
Key Mobile Device Reselection Algorithms
C1 = PSDavg - Dev_Rx_Rqdmin –max(Dev_Xmit_CCCHmax-Dev_Pwr_Class)
𝑪𝟐 = 𝑪𝟏 + 𝑪𝒆𝒍𝒍_𝑹𝒆𝒔𝒆𝒍𝒆𝒄𝒕_𝑶𝒇𝒇𝒔𝒆𝒕-max(Temp_Off_Penalty_Time)
21
RF Considerations
(Roaming)

Power Spectral Density Estimation
 Used by all devices cellular and WiFi devices to determine when
to roam. Utilize a standardized PSD estimate and use the
existing network stacks to perform “soft” handovers

Limitations: Distance, Shadow, and Rayleigh Fading.

Rayleigh fading can be addressed using orthogonal modulation
techniques
 Distance fading can be reduced by smaller cell sizes; however,
cannot be eliminated Note that the worst case distance for 802.11 is
~8400 meters (e.g. 802.11g)
𝑳𝒅 = 𝟏𝟎𝜸 𝒍𝒐𝒈 𝒅 , 𝟐 < 𝜸 < 𝟒 (dB, Urban environment)
 Shadow fading can be reduced by elevated antennas (e.g. dipole
antennas used to reduce impedance mismatch with 50ohm
connectors); however, cannot be fully eliminated.
𝑳𝒔 = 𝟏𝟎𝜸 𝒍𝒐𝒈 𝟒𝝅𝒅 𝝀 + 𝑳𝝈 , 𝑳𝝈 (dB) ~ N 𝟎, 𝝈
22
RF Considerations
(Roaming)

Power Spectral Density Estimation

Goal: Determine the best Power Spectral Density (PSD)
estimation method for the derivation of the PSD of sinusoid
signals in the presence of white noise utilizing different
estimation methods

Approach: Conduct a statistical experiment utilizing the
following different estimation methods: Blackman-Tukey
Correlogram, Welch Periodogram, Yule-Walker, Burg,
Covariance, Modified Covariance, and Multiple Signal
Classification (MUSIC).
23
RF Considerations
(Roaming)

Power Spectral Density Estimation (Continued)
 Approach: Conduct a statistical experiment following these
steps:
 Choose factors, levels, and ranges for the experiment
 Choose the response variables
 Choose the design of the experiment
 Conduct iterative experiments
 Perform JMP statistical analysis (using ANOVA)
 Based upon statistical analysis, determine conclusions and
recommendations
24
RF Considerations
(Roaming)

Power Spectral Density Estimation (Continued)
 Factors:
 Yule-Walker, Burg, Covariance, Modified Covariance, and
Multiple Signal Classification (MUSIC).
 Correlogram Ranges:
 Three different Lag levels: 10, 20, 70 point lag windows
(number of input points remained constant)

Periodogram Ranges:


Three different shift adjustments: 10, 20, 30 point shift
adjustments.
Autoregressive (AR) Ranges (Yule-Walker, Burg,
Covariance, Mod. Covariance, Music) :

Three different model orders M: 5, 15, 30
25
RF Considerations
(Roaming) (Non-Parametric Est.)

Blackman-Tukey (lag window) Correlogram


Results: In the experiment, increasing the lag window
increased the frequency resolution; however, note that in each
experiment M was much smaller than the total number of
points; therefore, the frequency resolution increased.
This demonstrated estimation noise, loss of resolution, and
spectral leakage; however, increasing the lag window L did in
fact improve spectral resolution, put it demonstrated the
tradeoff between spectral resolution and estimation noise.
𝑃𝐵𝑇 𝑒 𝑗Ω = 𝑅𝐵𝑇 𝑒 𝑗Ω =
𝐿
𝑚=−𝐿 𝑤
𝑚 𝑟𝑥𝑥 𝑚 𝑒 −𝑗Ω𝑚
26
RF Considerations
(Roaming) (Non-Parametric Est.)

Welch Periodogram


Results: In the experiment, increasing the overlap window
increased the frequency resolution; however, note that I was
not able to resolve the third frequency component. I kept the
number of sample points constant for all tests since a longer
observation does not reduce the variance.
This method introduced a tapered window (e.g. Hamming
window) prior to the DTFT and allowed for varying overlapping
frames (e.g. shift .10, 20, 30) or maximum, approx. 2/3 overlap
𝑟𝑥𝑥 𝑛 = 𝑤 𝑛 𝑥𝑜𝑣𝑒𝑟𝑙𝑎𝑝 𝑛
P 𝑓 =
𝑇
𝑁
𝑁−1
𝑛=0 𝑟𝑥𝑥
𝑛
2
−𝑗2𝜋𝑓𝑛𝑇
𝑒
27
RF Considerations
(Roaming) (Parametric Est.)

Yule-Walker

Results:By setting the parameter b0 = 1, the task is to estimate
ak. By utilizing higher order model resolutions (e.g. higher
model orders), the variance of the forward and backward linear
error prediction is minimized for higher model orders; therefore,
increased frequency resolution was achieved with higher
model orders. The all-pole model the affect of plot smoothing.
𝑃𝐴𝑅 𝑓 =

𝑇 ∗𝑣𝑎𝑟 𝑤
𝑝
1+ 𝑘=1 𝑎𝑝 𝑘 𝑒 −𝑗2𝜋𝑘𝑓𝑇 2
Burg

Results:Unlike the Yule-Walker, the Burg method does not
utilize the Autocorrelation matrix. Though accuracy improved
at higher model orders, the plots demonstrated spurious
frequency peaks at higher model orders.
28
RF Considerations
(Roaming) (Parametric Est.)

Covariance/Mod. Covariance


Results: Similar to the Yule-Walker method in the sense that it
is also an all-pole model. This model seeks to minimize the
forward error prediction with generating a non-singular event.
As the model order increased, frequency resolution improved.
This method demonstrated results very close to the Burg model
in the sense that spurious peaks were observed at the higher
model order
MUSIC

Results:Unlike the other Autoregressive models, the MUSIC
algorithm does carry the power spectral content of signals
forward. Therefore, it can only be used to identify spectral
content in frequency only, not magnitude.
29
RF Considerations
(Areas of Study)

Space-Time adaptive interference cancellation
 Based upon RF spectrum sensing, share RF measurement
information between base stations by Cell Controller
measurement unit


Canonical model coefficents communicated by Cell Controller related to
RF environment (e.g. neighboring cells)
Reduce adjacent channel power in adjacent cells

Areas of Further Study
 QR decomposition
 Back substitution
 Whitening beamforming

Innovative Software DSP Technology


Mindspeed
Coherent Logix
30
RF Considerations
(Conclusions)


Conclusion
 Based upon empirical JMP analysis results, the best
performing PSD estimation methods where the high
resolution: Blackman-Tukey Correlogram, Yule-Walker
(AR) method, and the Modified Covariance (AR) method.
 JMP analysis results indicated MUSIC scores high as well
for frequency estimation; however, it cannot be used for
PSD estimation.
Once PSD is estimated, the estimation coefficients (e.g.
matrices) can be sent back to the Cell Controller by base
stations.
 Adaptive filter can notch out interferers


Combination of LPF + HPF
Space-Time Adaptation techniques can be performed
31
Wireless Networks
Consolidated at Central Location
32
Design/IP Verification Test (DVT)
Functional/Performance Verification
33
Design Verification Model (DVM)
Functional/Performance Verification
34
Verify Network Processor (Si)
35
Conclusion

A new concept in regard to link layer communication as it relates to wireless
devices has been introduced that provides fully integrated support for the use of
802.11 and cellular protocol stacks
 A new hardware implementation (e.g. start of SoC design, functional
decomposition) had been introduced to support this new link layer idea that
wireless data frames are processed by hardware state machines (not by software).
 Mobile device issues were reviewed in regard to their state machine behavior,
capabilities, and typical interactions with existing wireless communication protocols.
I further described how a cell controller would support these devices as well as a
small research effort in regard to RF signal estimation and PSD derivation that
might be implemented on mobile devices on behalf of the cell controller (e.g.
network side RF measurement request). I also introduced how this might be used
to change the roaming paradigm of wireless devices (strictly based upon PSD using
“soft” handovers)
 Hardware design, development, and verification methodologies were introduced
that would be employed to develop such a SoC design as well as provide functional
verification after synthesis
36
Wired Technology
Switch/Router Behavior
37
Wired Technology
Switch/Router Behavior
38
WiFi Technology
Basic Frame Types (802.11a/b/g/n)





Beacon Transmissions/Probes
Probe Request
 ESS known if no Beacon
 Retrieved from Beacon if Beacon messages enabled
 ESS identified from other Probe Requests/Responses
Authentication
 Typically, Open Authentication Type
Association
 ESS/BSS combination
 DTIM interval, Association ID utilized for power save mask
 Maximum wireless clients
802.1x (Typically Extensible Authentication Protocol –EAP)
 Authentication (4-Way Handshake)
 Encryption
40
WiFi Technology
Basic Features



IP Address Domains
 Can be assigned per unique ESS/BSS combination
 Can be assigned per unique ESS
QoS
 Can be applied per ESS/BSS combination
 Can be applied per ESS combination
 Can be applied per IP domain
 Frames to RF (Weighted Fair Queue)
 Voice frames (Priority Queuing)/RED
Security
 WPA (PSK,AES) (over 802.1x)
 Handling of Broadcast Traffic

(i.e. Lowest Common Denominator for all devices associated)
41
WiFi Technology

WiFi Comparison

WiFi Direct Rough
equivalency to entire
IEEE 802.11 WiFi
features
 P2P Group Owner
(GO) acts like 802.11
Access Point

WiFi Direct Features
Same as WiFi
 QoS
 Security (WPA-2)
 Power Saving
42
WiFi Behavior
NAT Performed by AP
43
WiFi Behavior
NAT Performed by AP
44
Wireless Networks
IP Assigned External -AP Bridge Mode
45
Wireless Networks
Router External Performs NAT
46
Wireless Networks
IP Address Assigned External to AP
47
Wireless Networks
As Seen by Router -ARP cache
48
Wireless Networks
Roaming -AP = Bridge Mode
49
Wireless Networks
(Enterprise Class)
50
Wireless Networks
(VPN)
51
Wireless Networks
52
Wireless Networks
VPN -Mixed Environment
QoS Applied Across all Access Points
Based upon network configuration
53
Wireless Networks
54
Exemplar Frame Processing
55
Network Processor
(Example of Latest Technology)
56
Network Processor
(High Level Packet Walkthrough)
57
Network Processor
(High Level Packet Walkthrough)
58
Questions???
Thank You
59
HW Finite State Machines
Wireless
MAC Engine
Routing
Frame
Transmit/
Receive
HW
Queue
Frame
Classification
Encryption/
Decryption
Network
Buffer
Mgmt
Pattern
Matching
Network
Buffer
Mgmt
1…N
Core
Core
Core
L2
L2
Cache
L2Cache
Cache
I-Cache
I-Cache
I-Cache
L3
D-Cache
Cache
D-Cache
L3Cache
Cache
D-Cache L3
Coherency
System Memory
Mem
CTL
1…N
ADDR
ADDR
ADDR
ID
ID
ID
Key
Key
Info
KeyInfo
Info
IV
Info
IV
Info
IV Info
Mem.
Mem.
Cache
Mem.Cache
Cache
Instructions
Instructions
Instructions
Per
Flow & FSM
HW Cache
Structure
60
HW Finite State Machines
Wireless Link Layer
802.11/Cellular
Measure
Input Fifo
Network
Buffer
Mgmt
1…N
HW
Queue
Mem
CTL
I-Cache
I-Cache
I-Cache
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
Internal
DMA
System Memory
Core
Core
Core
L2
L2
Cache
L2Cache
Cache
Queue
Interface
Frame
Transmit/
Receive
Parse,
Classify
Output Fifo
802.11/Cellular
Roaming
L3
D-Cache
Cache
D-Cache
L3Cache
Cache
D-Cache L3
1…N
Cryptography
Frame Buffers
1…N
Coherency
IV Info
Key Info
Eth/VLAN Template
802.11 MAC Template
802.11 MGMT Subsys
Cellular MGMT
Subsys
802.11 and Cellular Subsystems
Handle Non-Real Time Link Layer Communication
Cellular MAC Template
Device State
Frame Queue Addr Lst
Instructions
Per
802.11 or Cellular
User Device
& FSM
HW Cache
Structure
61
HW Finite State Machines
Wireless Link Layer
802.11/Cellular
Measure
Input Fifo
Network
Buffer
Mgmt
1…N
HW
Queue
Mem
CTL
I-Cache
I-Cache
I-Cache
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
FrameProcessor
Mem. Cache
Internal
DMA
System Memory
Core
Core
Core
L2
L2
Cache
L2Cache
Cache
Queue
Interface
Frame
Transmit/
Receive
Parse,
Classify
Output Fifo
802.11/Cellular
Roaming
L3
D-Cache
Cache
D-Cache
L3Cache
Cache
D-Cache L3
1…N
Cryptography
Frame Buffers
1…N
Coherency
IV Info
Key Info
Eth/VLAN Template
1
802.11 MGMT Subsys
Cellular MGMT
Subsys
802.11 and Cellular Subsystems
Handle Non-Real Time Link Layer Communication
802.11 MAC Template
Cellular MAC Template
Device State
Frame Queue Addr Lst
Per
802.11 or Cellular
User Device
& FSM
HW Cache
Structure
2
62
Problem Overview

Goal
 Develop an advanced multi-nodal communication system that will
provide an integrated link layer for BOTH WiFi and Cellular protocols
within the same cell coverage areas.
 Approach
 Leverage existing WiFi ,Cellular and wired network protocol standards
 Leverage spatial diversity to extend range or seamlessly “roam” across
different radio cell sizes
 Centralize WiFi link layer communication
 Utilize existing network protocol standards to utilize banck-end
unreliable communication channels
 Integrate WiFi data path with Cellular data path subsystem
 Distribute WiFi link layer communication with backend system
 Create “Smart” WiFi & Cellular Access Points
 Communicate RF parametric information to backend system so
better network decisions can be made for individual devices
 Improve Power Spectral Density estimates so accurate information
is communicated to backend system
64
Problem Overview

Goal
 Develop an advanced multi-nodal communication system that will
provide an integrated link layer for BOTH WiFi and Cellular protocols
within the same cell coverage areas.
 Approach
 Leverage existing WiFi ,Cellular and wired network protocol standards
 Leverage spatial diversity to extend range or seamlessly “roam” across
different radio cell sizes
 Centralize WiFi link layer communication
 Utilize existing network protocol standards to utilize banck-end
unreliable communication channels
 Integrate WiFi data path with Cellular data path subsystem
 Distribute WiFi link layer communication with backend system
 Create “Smart” WiFi & Cellular Access Points
 Communicate RF parametric information to backend system so
better network decisions can be made for individual devices
 Improve Power Spectral Density estimates so accurate information
is communicated to backend system
65
Problem Overview

Goal
 Develop an advanced multi-nodal communication system that will
provide an integrated link layer for BOTH WiFi and Cellular protocols
within the same cell coverage areas.
 Approach
 Leverage existing WiFi ,Cellular and wired network protocol standards
 Leverage spatial diversity to extend range or seamlessly “roam” across
different radio cell sizes
 Centralize WiFi link layer communication
 Utilize existing network protocol standards to utilize banck-end
unreliable communication channels
 Integrate WiFi data path with Cellular data path subsystem
 Distribute WiFi link layer communication with backend system
 Create “Smart” WiFi & Cellular Access Points
 Communicate RF parametric information to backend system so
better network decisions can be made for individual devices
 Improve Power Spectral Density estimates so accurate information
is communicated to backend system
66
Communication Techniques

Problem Statements


WiFi link layer does not support
roaming efficiently across multiple
access points on the network side.
WiFi and Cellular technologies continue
to converge.




Performance at the Network “Edge”
continues to increase




“Smart” phones support both WiFi and
Cellular communication
Link layer communication between the
two technologies have not converged in
the market space [yet]
Cell sizes continue to shrink to support
large capacity
Roaming needs to be high performance
between [Wi-Fi & Cellular] Access
Points
WiFi roaming at link layer is not well
distributed
Reduce number of frames dropped
during a roaming event
How to implement


Not impact existing standards (.e.g.
802.11 & UMTS) in the market place
Provide ubiquitous wireless coverage
67
Communication Techniques

Problem Statements


WiFi link layer does not support roaming efficiently across multiple access points
on the network side.
WiFi and Cellular technologies continue to converge.




Performance at the Network “Edge” continues to increase




“Smart” phones support both WiFi and Cellular communication
Link layer communication between the two technologies have not converged in the market
space [yet]
Cell sizes continue to shrink to support large capacity
Roaming needs to be high performance between [Wi-Fi & Cellular] Access Points
WiFi roaming at link layer is not well distributed
Reduce number of frames dropped during a roaming event
How to implement


Not impact existing standards (.e.g. 802.11 & UMTS) in the market place
Provide ubiquitous wireless coverage across both technologies
68
69
OSI Model
Link Layer