CHANNEL ERROR COMPENSATOR

At the beginning of each frame, the compensation block classifies a flow (i.e., a CID) as sensing a good or a bad channel on the basis of the CINR reports received by the relevant target SS. Accordingly, CIDs destined to an SS sensing a lossy channel are marked as “banned” from being transmitted. When the error-free service scheduler selects a head-of-line (HOL) packet to be extracted from queue i, it interacts with the compensator to check whether the CID of the selected packet is banned or not. If it is banned (the packet would risk to be corrupted if transmitted over the radio link), then the scheduler looks for a packet belonging to an unmarked flow of the same class (i.e., in the same queue) to transmit instead of the HOL packet. To help the scheduler in selecting the substitute flow, the compensator manages a debit/credit counter for each active (leading or lagging) flow. Each time a flow substitutes another one for transmission, it is considered as a leading flow, and its credit counter increases by one; at the same time, the replaced flow is considered as a lagging flow, and its debit counter increases by one. Each time a HOL packet cannot be transmitted due to poor channel conditions, the substitute flow is the one with the highest debit counter among unmarked flows of the same class. If no unmarked packets can be found in the queue, then the entire queue is banned and the turn passes to the successive queue.

The pseudocode listed below briefly shows how the compensator works and interacts with the scheduler, regardless of the specific algorithm (WF²Q+ or DRR) this latter uses.

1 ban flow if CINR > CINRthr;

2 until available resources {

3       until a pkt is dequed || all queues examined {

4               scheduler selects queue x : x !banned i.e. scheduler selects flow i = head(x);

5               if i is marked {

6                      select less serviced flow j : j !banned;

7                      if !found {

8                                    mark x;

9                                    goto 3; }

10                    else {

11                                   deque pkt;

12                                   modify scheduler variables;

13                                   modify credit/debit counter of i and j;

14                                   exit; } }

15              else {

16                     modify scheduler variables;

17                     deque pkt; } } }

At the beginning of each frame, the compensator bans the flows directed to SSs, which currently sense a lossy channel (line 1), according to the received channel state information reports. For the duration of the DL subframe, packets within the queues are examined by the scheduler until at least one queue contains a packet belonging to an unbanned CID (lines 2–3). The scheduler selects the queue to be served among all backlogged (and unbanned) queues; this corresponds to select the first packet in the queue (line 4). If this packet belongs to an unbanned flow, then it gains service and the variables for the specific scheduling algorithm (WF²Q+ or DRR) are updated accordingly (lines 15–17). These variables can be the quantum and deficit counter values for DRR, and (S, F, V) values for WF²Q+. Otherwise, if the HOL packet is banned, the scheduler starts searching for the unbanned flow (in the same queue) that has received the worst service in the past (lines 5–6). If no unbanned flow exists, the Scheduler considers the entire queue as no longer servable and selects another queue among the backlogged ones, if any (lines 7–9). If at least an unbanned flow exists, the first enqueued packet belonging to this flow is selected and the scheduling variables are updated accordingly (lines 10–14). Also credit/debit counters for the originally selected flow and the substitute flow are consequently updated.

BANDWIDTH REQUESTS AND CINR REPORTS | QoS and Fairness in WiMAX

To make the channel-aware scheduler effectively operate, information about the quality of the wireless links toward each SS must be gathered by the BS. To this aim, the wireless channel state is continuously monitored by each SS, which gets CINR (carrier to interference and noise ratio) measurements, updates estimates of the CINR’s mean and standard deviation values, and reports them back to the BS through specific signalling messages provided by the IEEE 802.16 standard, e.g., through BR messages (normally used by SSs for requesting bandwidth), which can also include the CINR report.

The compensation block is in charge of collecting received information on the channel qualities and making the BS aware of the status of each link toward the SSs. Two assumptions hold in our reference scenario:

§  Wireless channel quality of each connection changes on a per-frame basis, but it remains constant for a frame duration; i.e, the BS assumes that the CINR report received by the SS at the frame beginning is valid for the entire subsequent frame duration. This is equivalent to modelling a block-fading channel, which is a feasible assumption for the slowly varying radio links of the fixed WiMAX PMP scenario.

§  Perfect channel state information is available at the BS. This means that the CINR reports must be correctly received by the BS over an error-free feedback channel.

The monitored channel quality at the SS is compared against the allowed values of signal-to-noise ratio and receiver sensitivity for each transport mode (burst profile) at a given bit error rate (BER). As soon as the received power level becomes lower than the receiver sensitivity threshold for a given DL transport mode and target BER, then a more robust profile must be chosen. This change in the transport mode can proceed until the most robust modulation/coding scheme (BPSK modulation with 1/2 coding rate) is chosen. A channel is considered “good” as long as a more robust transport mode can be selected, which is capable to cope with the channel adverse conditions. Otherwise, the channel is finally considered as “bad.”

PACKET CLASSIFIER | QoS and Fairness in WiMAX

Packets received by the BS and destined to the DL are sorted by the packet classifier and buffered into one of the per-class queues in the BS, according to the class of service they belong to. In the queues, packets wait for transmission over the air toward a given SS. Queues are managed by the scheduler on a frame basis.

The main task of the classifier is to match the CID of the incoming packets with the traffic class they belong to, and to queue such packets in their relevant buffers. Only the service received by rtPS, nrtPS, and BE queues will be analyzed, because flows belonging to the UGS and ertPS service classes are assumed to receive a constant reserved amount of bandwidth frame-by-frame and not need to be scheduled. This choice allows high demanding (UGS and ertPS) flows to save up the time required for the BR process.

Another important task of the classifier is to timestamp each incoming packet according to its arrival time. This information is exploited by the buffer manager to recognize when a queued packet’s timeout expires, i.e., when its elapsed waiting time exceeds the maximum tolerated latency.