Basic model

M/M/1 Queue
M/M/C Queue Erlang-C model
PS models nicely capture the bandwidth sharing and multiplexing effect of TCP at the flow-level
However, RTT not taken into account On a smaller time scale the packet-level model describes more detailed events.

M/G/1 PS for single class:
M/G/1 PS for multi-class: product-form solution
what product-form means +relation to HSDPA

GSM –(Erlang-B model)

use a combination of FDMA and TDMA as multiple access scheme.

TDMA

devides the radio spectrum into time slots
pro:user can use complete frequency band
cons: no continuous transmission

FDMA

allows multiple users to send data through a single communication channel, by dividing the bandwidth of the channel into seperate non-overlapping frequency sub-channels.
pros:continuous transmission
con: use can only use a single frequency

CDMA

pro:high data rates+continuous transmission+easy network expansion
cons:interference and near-far problem

GPRS –(combined blocking/PS model)

HSCSD –(multi-rate.knapsack model)

close-form solution

Closed-form expressions are often beyond reach because the population-size constraint complicates the evaluation of terms summing over all possible queue-length vectors, which appear in for instance the normalization constant.

product-form

Networks with product form solutions
In this class of networks, in spite of the high level of interaction between the nodes, the joint distribution of all the nodes is the product of the marginal distributions of the individual nodes. Roughly speaking, it implies that the stationary distribution of the network can be obtained by multiplying the stationary distributions of the individual nodes assuming that each node is in isolation and subject to Poisson arrivals.

kaufman-roberts recursion

Since the multi-dimensional state psace can be mapped into a one-dimensional state space without influencing the resulting blocking probabilities.
The KR recursion provides a simple and efficient method for calculating steady-state probabilities without the need for solving complex linear algebraic equations.

Compared to PF
larger and more realistic numbers of traffic classes could be investigated without running into numerical difficulties.

WLAN 802.11 –(Bianchi model)

The general step:
A station with a new packet to transmit monitors the channel activity. If the channel is idle for a period of time equal to a distributed interframe space(DIFS),the station transmits.
At this point, the station generates a random backoff interval before transmitting. At each packet transmission,the backoff time is uniformly chosen in the range $(0,w-1)$. The value $w$ is called contention windows.
The ACk is immediately transmitted at the end of the packet,after a period of time called short interframe space(SIFS). As the SIFS(plus the propagation delay) is shorter than a DIFS, no other station is able to detect the channel idle for a DIFS until the end of the ACK.

• contention windows At the first attempt,$w$ is set equal to value $CW_min$. After each unsuccessful transmission, $ w $ is doubled,up to a maximum value$ CW_{max}=x^m{CW}_min $

distributed coordination function (DCF)

the fundamental mechanism to access the medium is called distributed coordination function. This is a random access scheme,based on the carrier sense multiple access with collided avoidance(CSMA/CA).

DCF describes two techniques to employ for packet transmission.

• two-way handshaking

• RTS/CTS Before transmitting a packet, a station operating in RTS/CTS modereserves the channel by sending a special Request-To-Sent short frame. The destination station acknowledges the receipt of an RTS frame by sending back a Clear-To-Send frame,after which normal packet transmission and ACK response ouucrs. Since collision may occur only on the RTS frame, and it is detected by the lack of CTS response benefit:Increase the system performance by reducing the duration of a collision when long messages are transmitted.

  SINR信干燥比

  SINR is commonly used in wireless communication as a way to measure the quality of wireless connection ** the SINR is decreasing with the distance **

  Multipath fading

  On a small time scale,multiple copies of a given signal may strengthen or weaken one another
  These different copies have different path lengths,hence,they arrive with different phases at the receiver.

## UMTS(network of loss) parallel transmissions

HSDPA

shared channel

packet scheduling

opportunistic schedule: assign the resources to the mobile terminals having the best channel conditions by taking advantage of multi0user diversity
Although it delivers the best overall system throughput, terminals with bad channel conditions over an extended period of time experience significant QoS degradation as they are less likely to be scheduled 1This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible at all.

round robin

Maximum SNR

always serve flow with best SINR

Proportional Fairness

Trying to maximize the total throughput of the network (wired or not) while at the same time allowing all users at least a minimal level of service.

always serve flow with best relative SINR

Product-form solution via multi-calss PS model

channel quality indicator

很多MIMO技术 需要用到信道质量指示器来反馈信道特征给发射器，用以调整发射天线的数据速率，实现自适应调制 ## Long- Term Evolution

OFDMA

### MIMO 多径衰落是影响通信质量的主要因素，但MIMO系统能有效利用多径的影响来提高系统容量

5G

-new communication paradigms Combining cellular and ad-hoc type of communications(D2D)

5G beamforming pico cells frequency reuse

Spectrum reuse
exploit new spectrum
与之前的蜂窝网络（如3G和4G）相比，频率越高、波长越短、衍射能力越弱，因此在遇到有障碍物的地方，其辐射范围会变得很小

device to device communications:increase spectrum efficiency.

QoS in communication networks

two approaches:

QoS Control

Bandwidth(over)provisioning

perform regular,coarse traffic measurements
use rules of thumb
no QoS control mechanisms

traffic policying

The differences between traffic policing and traffic shaping are as follows:

Traffic policing directly discards the packets whose rate exceeds the rate limit. Traffic shaping, however, buffers the packets whose rate is greater than the traffic shaping rate. When there are sufficient tokens in the token bucket, the device forwards buffered packets at an even rate. Traffic shaping increases the delay, whereas traffic policing does not.