• 沒有找到結果。

N/A
N/A
Protected

Copied!
23
0
0

(1)

(2)

Multipath Propagation

2

RX just sums up all Multi Path Component (MPC).

(3)

Multipath Channel Impulse Response

t0

t

0

1

2

3

4

5

6 t(t

0)

b

(t,t)

An example of the time-varying discrete-time impulse response for a multipath radio channel

The channel impulse response when 𝑡 = 𝑡# (what you receive at the receiver when you send an impulse at time 𝑡#)

𝜏# = 0, and represents the time the first signal arrives at the receiver.

Summed signal of all multipath components arriving at 𝜏'~𝜏')*.

Excess delay: the delay with respect to the first arriving signal (𝜏)

Maximum excess delay: the delay of latest arriving signal

(4)

t

t0

0

1

2

3

4

5

6 t(t

0) t(t1) t1

t2

t(t2) t3

t(t3)

b

(t,t)

Because the transmitter, the receiver, or the reflectors are moving, the impulse response is time-variant.

(5)

Multipath Channel Impulse Response

-

=

1

0

N

i

i i

i c i

b

-

=

1

0

N i

i i

i

b

h t q d t t

Phase change due to different arriving time Additional phase change due to reflections

Amplitude change (mainly path loss) Summation over all MPC

(6)

t

t0

0

1

2

3

4

5

6 t(t

0) t(t1) t1

t2

t(t2) t3

t(t3)

b

(t,t)

Following this axis, we study how “spread-out” the impulse response are.

(related to the physical layout of the TX, the RX, and the reflectors at a single time point)

(7)

t

t0

0

1

2

3

4

5

6 t(t

0) t(t1) t1

t2

t(t2) t3

t(t3)

b

(t,t)

Following this axis, we study how “spread-out” the impulse response are.

(related to the physical layout of the TX, the RX, and the reflectors at a single time point)

(8)

Power delay profile

b

B

2

5

𝑡; 𝜏

6

TX 𝑝(𝑡) RX

Average over

several measurements in a local area

(9)

Example: power delay profile

9

From a 900 MHz cellular system in San Francisco

(10)

Example: power delay profile

10

Inside a grocery store at 4 GHz

(11)

(12)

Time dispersion parameters

k

k k

k k

k k k

k k

2 2 __

t

First moment of the power delay profile

k

k k

k k

k k k

k k

2

2 2 2 __2

__ 2 __

2

t

= -

Second moment of the power delay profile Square root of the second central moment of the power delay

profile

(13)

(14)

(15)

(16)

Coherence Bandwidth

t

c

t

» 1 B

c

Those two are approximations derived from empirical results.

(17)

(18)

Coherence Bandwidth

18

f t

Transmitted Signal

𝑇8

𝑇8: symbol period

𝐵8 𝐵8: signal bandwidth

t0

0

1

2

3

4

5

6

𝑇8 ≈ 1 𝐵8

𝐵<

𝜎>

(19)

𝐵<

𝐵8 f

f

8

<

TX signal

Channel

RX signal

×

= =∗

t0

0

1

2

3

4

5

6

𝜎>

𝑇8 t

8

< 𝜎

>

These will become inter- symbol interference!

𝑇8

(20)

𝐵<

f 𝐵8

f

8

<

TX signal

Channel

RX signal

×

= =∗

8

>

t0

0

1

2

3

4

5

6

𝑃(𝑡; 𝜏)

𝜎>

t 𝑇8

𝑇8 No significant ISI

(21)

(22)

Example

0 1 2 3 4 5 -30dB

-20dB -10dB 0dB

t P(t)

Would this channel be suitable for AMPS or GSM without the use of an equalizer?

P s P

k

k k

k

k

__

2 2

2 2

2 2

__

2

k

k k

k

k

(23)

Example

C

C

t

__ 2 2

__2

C

s

t

• When paging in from disk, we need a free frame of physical memory to hold the data we’re reading in. • In reality, size of physical memory is

One, the response speed of stock return for the companies with high revenue growth rate is leading to the response speed of stock return the companies with

• cost-sensitive classifier: low cost but high error rate. • traditional classifier: low error rate but

An OFDM signal offers an advantage in a channel that has a frequency selective fading response.. As we can see, when we lay an OFDM signal spectrum against the

“Big data is high-volume, high-velocity and high-variety information assets that demand cost-effective, innovative forms of information processing for enhanced?. insight and

For the data sets used in this thesis we find that F-score performs well when the number of features is large, and for small data the two methods using the gradient of the

• When the coherence bandwidth is low, but we need to use high data rate (high signal bandwidth). • Channel is unknown

If we would like to use both training and validation data to predict the unknown scores, we can record the number of iterations in Algorithm 2 when using the training/validation